SRX10201273 GSM5121509 SRP308746 SRS8347755 GSM5121509 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121509 GSM5121509 GEO Accession GSM5121509 SRX10201274 GSM5121510 SRP308746 SRS8347753 GSM5121510 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121510 GSM5121510 GEO Accession GSM5121510 SRX10201275 GSM5121511 SRP308746 SRS8347754 GSM5121511 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121511 GSM5121511 GEO Accession GSM5121511 SRX10201276 GSM5121512 SRP308746 SRS8347756 GSM5121512 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121512 GSM5121512 GEO Accession GSM5121512 SRX10201277 GSM5121513 SRP308746 SRS8347760 GSM5121513 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121513 GSM5121513 GEO Accession GSM5121513 SRX10201278 GSM5121514 SRP308746 SRS8347757 GSM5121514 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121514 GSM5121514 GEO Accession GSM5121514 SRX10201279 GSM5121515 SRP308746 SRS8347759 GSM5121515 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121515 GSM5121515 GEO Accession GSM5121515 SRX10201280 GSM5121516 SRP308746 SRS8347758 GSM5121516 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121516 GSM5121516 GEO Accession GSM5121516 SRX10201281 GSM5121517 SRP308746 SRS8347761 GSM5121517 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121517 GSM5121517 GEO Accession GSM5121517 SRX10201282 GSM5121518 SRP308746 SRS8347763 GSM5121518 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121518 GSM5121518 GEO Accession GSM5121518 SRX10201283 GSM5121519 SRP308746 SRS8347765 GSM5121519 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121519 GSM5121519 GEO Accession GSM5121519 SRX10201284 GSM5121520 SRP308746 SRS8347762 GSM5121520 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121520 GSM5121520 GEO Accession GSM5121520 SRX10201285 GSM5121521 SRP308746 SRS8347764 GSM5121521 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121521 GSM5121521 GEO Accession GSM5121521 SRX10201286 GSM5121522 SRP308746 SRS8347766 GSM5121522 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121522 GSM5121522 GEO Accession GSM5121522 SRX10201287 GSM5121523 SRP308746 SRS8347767 GSM5121523 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121523 GSM5121523 GEO Accession GSM5121523 SRX10201288 GSM5121524 SRP308746 SRS8347768 GSM5121524 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121524 GSM5121524 GEO Accession GSM5121524 SRX10201289 GSM5121525 SRP308746 SRS8347769 GSM5121525 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121525 GSM5121525 GEO Accession GSM5121525 SRX10201290 GSM5121526 SRP308746 SRS8347770 GSM5121526 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121526 GSM5121526 GEO Accession GSM5121526 SRX10201291 GSM5121527 SRP308746 SRS8347771 GSM5121527 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121527 GSM5121527 GEO Accession GSM5121527 SRX10201292 GSM5121528 SRP308746 SRS8347772 GSM5121528 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121528 GSM5121528 GEO Accession GSM5121528 SRX10201293 GSM5121529 SRP308746 SRS8347773 GSM5121529 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121529 GSM5121529 GEO Accession GSM5121529 SRX10201294 GSM5121530 SRP308746 SRS8347774 GSM5121530 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121530 GSM5121530 GEO Accession GSM5121530 SRX10201295 GSM5121531 SRP308746 SRS8347776 GSM5121531 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121531 GSM5121531 GEO Accession GSM5121531 SRX10201296 GSM5121532 SRP308746 SRS8347775 GSM5121532 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121532 GSM5121532 GEO Accession GSM5121532 SRX10201297 GSM5121533 SRP308746 SRS8347777 GSM5121533 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121533 GSM5121533 GEO Accession GSM5121533 SRX10201298 GSM5121534 SRP308746 SRS8347778 GSM5121534 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121534 GSM5121534 GEO Accession GSM5121534 SRX10201299 GSM5121535 SRP308746 SRS8347779 GSM5121535 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121535 GSM5121535 GEO Accession GSM5121535 SRX10201300 GSM5121536 SRP308746 SRS8347780 GSM5121536 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121536 GSM5121536 GEO Accession GSM5121536 SRX10201301 GSM5121537 SRP308746 SRS8347782 GSM5121537 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121537 GSM5121537 GEO Accession GSM5121537 SRX10201302 GSM5121538 SRP308746 SRS8347784 GSM5121538 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121538 GSM5121538 GEO Accession GSM5121538 SRX10201303 GSM5121539 SRP308746 SRS8347781 GSM5121539 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121539 GSM5121539 GEO Accession GSM5121539 SRX10201304 GSM5121540 SRP308746 SRS8347783 GSM5121540 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121540 GSM5121540 GEO Accession GSM5121540 SRX10201305 GSM5121541 SRP308746 SRS8347785 GSM5121541 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121541 GSM5121541 GEO Accession GSM5121541 SRX10201306 GSM5121542 SRP308746 SRS8347788 GSM5121542 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121542 GSM5121542 GEO Accession GSM5121542 SRX10201307 GSM5121543 SRP308746 SRS8347786 GSM5121543 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121543 GSM5121543 GEO Accession GSM5121543 SRX10201308 GSM5121544 SRP308746 SRS8347787 GSM5121544 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121544 GSM5121544 GEO Accession GSM5121544 SRX10201309 GSM5121545 SRP308746 SRS8347789 GSM5121545 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121545 GSM5121545 GEO Accession GSM5121545 SRX10201310 GSM5121546 SRP308746 SRS8347790 GSM5121546 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121546 GSM5121546 GEO Accession GSM5121546 SRX10201311 GSM5121547 SRP308746 SRS8347792 GSM5121547 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121547 GSM5121547 GEO Accession GSM5121547 SRX10201312 GSM5121548 SRP308746 SRS8347791 GSM5121548 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121548 GSM5121548 GEO Accession GSM5121548 SRX10201313 GSM5121549 SRP308746 SRS8347795 GSM5121549 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121549 GSM5121549 GEO Accession GSM5121549 SRX10201314 GSM5121550 SRP308746 SRS8347793 GSM5121550 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121550 GSM5121550 GEO Accession GSM5121550 SRX10201315 GSM5121551 SRP308746 SRS8347794 GSM5121551 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121551 GSM5121551 GEO Accession GSM5121551 SRX10201316 GSM5121552 SRP308746 SRS8347796 GSM5121552 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121552 GSM5121552 GEO Accession GSM5121552 SRX10201317 GSM5121553 SRP308746 SRS8347797 GSM5121553 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121553 GSM5121553 GEO Accession GSM5121553 SRX10201318 GSM5121554 SRP308746 SRS8347798 GSM5121554 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121554 GSM5121554 GEO Accession GSM5121554 SRX10201319 GSM5121555 SRP308746 SRS8347799 GSM5121555 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121555 GSM5121555 GEO Accession GSM5121555 SRX10201320 GSM5121556 SRP308746 SRS8347800 GSM5121556 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121556 GSM5121556 GEO Accession GSM5121556 SRX10201321 GSM5121557 SRP308746 SRS8347801 GSM5121557 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121557 GSM5121557 GEO Accession GSM5121557 SRX10201322 GSM5121558 SRP308746 SRS8347802 GSM5121558 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121558 GSM5121558 GEO Accession GSM5121558 SRX10201323 GSM5121559 SRP308746 SRS8347803 GSM5121559 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121559 GSM5121559 GEO Accession GSM5121559 SRX10201324 GSM5121560 SRP308746 SRS8347804 GSM5121560 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121560 GSM5121560 GEO Accession GSM5121560 SRX10201325 GSM5121561 SRP308746 SRS8347805 GSM5121561 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121561 GSM5121561 GEO Accession GSM5121561 SRX10201326 GSM5121562 SRP308746 SRS8347806 GSM5121562 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121562 GSM5121562 GEO Accession GSM5121562 SRX10201327 GSM5121563 SRP308746 SRS8347807 GSM5121563 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121563 GSM5121563 GEO Accession GSM5121563 SRX10201328 GSM5121564 SRP308746 SRS8347808 GSM5121564 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121564 GSM5121564 GEO Accession GSM5121564 SRX10201329 GSM5121565 SRP308746 SRS8347809 GSM5121565 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121565 GSM5121565 GEO Accession GSM5121565 SRX10201330 GSM5121566 SRP308746 SRS8347810 GSM5121566 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121566 GSM5121566 GEO Accession GSM5121566 SRX10201331 GSM5121567 SRP308746 SRS8347811 GSM5121567 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121567 GSM5121567 GEO Accession GSM5121567 SRX10201332 GSM5121568 SRP308746 SRS8347812 GSM5121568 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121568 GSM5121568 GEO Accession GSM5121568 SRX10201333 GSM5121569 SRP308746 SRS8347813 GSM5121569 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121569 GSM5121569 GEO Accession GSM5121569 SRX10201334 GSM5121570 SRP308746 SRS8347814 GSM5121570 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121570 GSM5121570 GEO Accession GSM5121570 SRX10201335 GSM5121571 SRP308746 SRS8347815 GSM5121571 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121571 GSM5121571 GEO Accession GSM5121571 SRX10201336 GSM5121572 SRP308746 SRS8347816 GSM5121572 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121572 GSM5121572 GEO Accession GSM5121572 SRX10201337 GSM5121573 SRP308746 SRS8347817 GSM5121573 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121573 GSM5121573 GEO Accession GSM5121573 SRX10201338 GSM5121574 SRP308746 SRS8347818 GSM5121574 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121574 GSM5121574 GEO Accession GSM5121574 SRX10201339 GSM5121575 SRP308746 SRS8347819 GSM5121575 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121575 GSM5121575 GEO Accession GSM5121575 SRX10201340 GSM5121576 SRP308746 SRS8347820 GSM5121576 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121576 GSM5121576 GEO Accession GSM5121576 SRX10201341 GSM5121577 SRP308746 SRS8347821 GSM5121577 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121577 GSM5121577 GEO Accession GSM5121577 SRX10201342 GSM5121578 SRP308746 SRS8347822 GSM5121578 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121578 GSM5121578 GEO Accession GSM5121578 SRX10201343 GSM5121579 SRP308746 SRS8347823 GSM5121579 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121579 GSM5121579 GEO Accession GSM5121579 SRX10201344 GSM5121580 SRP308746 SRS8347824 GSM5121580 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121580 GSM5121580 GEO Accession GSM5121580 SRX10201345 GSM5121581 SRP308746 SRS8347826 GSM5121581 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121581 GSM5121581 GEO Accession GSM5121581 SRX10201346 GSM5121582 SRP308746 SRS8347825 GSM5121582 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121582 GSM5121582 GEO Accession GSM5121582 SRX10201347 GSM5121583 SRP308746 SRS8347827 GSM5121583 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121583 GSM5121583 GEO Accession GSM5121583 SRX10201348 GSM5121584 SRP308746 SRS8347828 GSM5121584 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121584 GSM5121584 GEO Accession GSM5121584 SRX10201349 GSM5121585 SRP308746 SRS8347829 GSM5121585 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121585 GSM5121585 GEO Accession GSM5121585 SRX10201350 GSM5121586 SRP308746 SRS8347830 GSM5121586 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121586 GSM5121586 GEO Accession GSM5121586 SRX10201351 GSM5121587 SRP308746 SRS8347831 GSM5121587 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121587 GSM5121587 GEO Accession GSM5121587 SRX10201352 GSM5121588 SRP308746 SRS8347832 GSM5121588 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121588 GSM5121588 GEO Accession GSM5121588 SRX10201353 GSM5121589 SRP308746 SRS8347834 GSM5121589 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121589 GSM5121589 GEO Accession GSM5121589 SRX10201354 GSM5121590 SRP308746 SRS8347833 GSM5121590 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121590 GSM5121590 GEO Accession GSM5121590 SRX10201355 GSM5121591 SRP308746 SRS8347835 GSM5121591 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121591 GSM5121591 GEO Accession GSM5121591 SRX10201356 GSM5121592 SRP308746 SRS8347838 GSM5121592 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121592 GSM5121592 GEO Accession GSM5121592 SRX10201357 GSM5121593 SRP308746 SRS8347836 GSM5121593 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121593 GSM5121593 GEO Accession GSM5121593 SRX10201358 GSM5121594 SRP308746 SRS8347837 GSM5121594 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121594 GSM5121594 GEO Accession GSM5121594 SRX10201359 GSM5121595 SRP308746 SRS8347840 GSM5121595 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121595 GSM5121595 GEO Accession GSM5121595 SRX10201360 GSM5121596 SRP308746 SRS8347839 GSM5121596 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121596 GSM5121596 GEO Accession GSM5121596 SRX10201361 GSM5121597 SRP308746 SRS8347841 GSM5121597 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121597 GSM5121597 GEO Accession GSM5121597 SRX10201362 GSM5121598 SRP308746 SRS8347842 GSM5121598 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121598 GSM5121598 GEO Accession GSM5121598 SRX10201363 GSM5121599 SRP308746 SRS8347843 GSM5121599 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121599 GSM5121599 GEO Accession GSM5121599 SRX10201364 GSM5121600 SRP308746 SRS8347844 GSM5121600 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121600 GSM5121600 GEO Accession GSM5121600 SRX10201365 GSM5121601 SRP308746 SRS8347845 GSM5121601 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121601 GSM5121601 GEO Accession GSM5121601 SRX10201366 GSM5121602 SRP308746 SRS8347846 GSM5121602 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121602 GSM5121602 GEO Accession GSM5121602 SRX10201367 GSM5121603 SRP308746 SRS8347847 GSM5121603 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121603 GSM5121603 GEO Accession GSM5121603 SRX10201368 GSM5121604 SRP308746 SRS8347848 GSM5121604 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121604 GSM5121604 GEO Accession GSM5121604 SRX10201369 GSM5121605 SRP308746 SRS8347849 GSM5121605 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121605 GSM5121605 GEO Accession GSM5121605 SRX10201370 GSM5121606 SRP308746 SRS8347851 GSM5121606 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121606 GSM5121606 GEO Accession GSM5121606 SRX10201371 GSM5121607 SRP308746 SRS8347850 GSM5121607 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121607 GSM5121607 GEO Accession GSM5121607 SRX10201372 GSM5121608 SRP308746 SRS8347852 GSM5121608 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121608 GSM5121608 GEO Accession GSM5121608 SRX10201373 GSM5121609 SRP308746 SRS8347853 GSM5121609 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121609 GSM5121609 GEO Accession GSM5121609 SRX10201374 GSM5121610 SRP308746 SRS8347857 GSM5121610 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121610 GSM5121610 GEO Accession GSM5121610 SRX10201375 GSM5121611 SRP308746 SRS8347855 GSM5121611 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121611 GSM5121611 GEO Accession GSM5121611 SRX10201376 GSM5121612 SRP308746 SRS8347854 GSM5121612 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121612 GSM5121612 GEO Accession GSM5121612 SRX10201377 GSM5121613 SRP308746 SRS8347856 GSM5121613 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121613 GSM5121613 GEO Accession GSM5121613 SRX10201378 GSM5121614 SRP308746 SRS8347859 GSM5121614 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121614 GSM5121614 GEO Accession GSM5121614 SRX10201379 GSM5121615 SRP308746 SRS8347858 GSM5121615 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121615 GSM5121615 GEO Accession GSM5121615 SRX10201380 GSM5121616 SRP308746 SRS8347860 GSM5121616 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121616 GSM5121616 GEO Accession GSM5121616 SRX10201381 GSM5121617 SRP308746 SRS8347861 GSM5121617 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121617 GSM5121617 GEO Accession GSM5121617 SRX10201382 GSM5121618 SRP308746 SRS8347863 GSM5121618 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121618 GSM5121618 GEO Accession GSM5121618 SRX10201383 GSM5121619 SRP308746 SRS8347862 GSM5121619 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121619 GSM5121619 GEO Accession GSM5121619 SRX10201384 GSM5121620 SRP308746 SRS8347864 GSM5121620 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121620 GSM5121620 GEO Accession GSM5121620 SRX10201385 GSM5121621 SRP308746 SRS8347865 GSM5121621 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121621 GSM5121621 GEO Accession GSM5121621 SRX10201386 GSM5121622 SRP308746 SRS8347867 GSM5121622 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121622 GSM5121622 GEO Accession GSM5121622 SRX10201387 GSM5121623 SRP308746 SRS8347866 GSM5121623 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121623 GSM5121623 GEO Accession GSM5121623 SRX10201388 GSM5121624 SRP308746 SRS8347870 GSM5121624 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121624 GSM5121624 GEO Accession GSM5121624 SRX10201389 GSM5121625 SRP308746 SRS8347868 GSM5121625 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121625 GSM5121625 GEO Accession GSM5121625 SRX10201390 GSM5121626 SRP308746 SRS8347871 GSM5121626 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121626 GSM5121626 GEO Accession GSM5121626 SRX10201391 GSM5121627 SRP308746 SRS8347869 GSM5121627 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121627 GSM5121627 GEO Accession GSM5121627 SRX10201392 GSM5121628 SRP308746 SRS8347872 GSM5121628 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121628 GSM5121628 GEO Accession GSM5121628 SRX10201393 GSM5121629 SRP308746 SRS8347873 GSM5121629 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121629 GSM5121629 GEO Accession GSM5121629 SRX10201394 GSM5121630 SRP308746 SRS8347874 GSM5121630 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121630 GSM5121630 GEO Accession GSM5121630 SRX10201395 GSM5121631 SRP308746 SRS8347875 GSM5121631 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121631 GSM5121631 GEO Accession GSM5121631 SRX10201396 GSM5121632 SRP308746 SRS8347876 GSM5121632 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121632 GSM5121632 GEO Accession GSM5121632 SRX10201397 GSM5121633 SRP308746 SRS8347877 GSM5121633 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121633 GSM5121633 GEO Accession GSM5121633 SRX10201398 GSM5121634 SRP308746 SRS8347878 GSM5121634 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121634 GSM5121634 GEO Accession GSM5121634 SRX10201399 GSM5121635 SRP308746 SRS8347879 GSM5121635 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121635 GSM5121635 GEO Accession GSM5121635 SRX10201400 GSM5121636 SRP308746 SRS8347880 GSM5121636 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121636 GSM5121636 GEO Accession GSM5121636 SRX10201401 GSM5121637 SRP308746 SRS8347881 GSM5121637 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121637 GSM5121637 GEO Accession GSM5121637 SRX10201402 GSM5121638 SRP308746 SRS8347882 GSM5121638 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121638 GSM5121638 GEO Accession GSM5121638 SRX10201403 GSM5121639 SRP308746 SRS8347883 GSM5121639 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121639 GSM5121639 GEO Accession GSM5121639 SRX10201404 GSM5121640 SRP308746 SRS8347884 GSM5121640 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121640 GSM5121640 GEO Accession GSM5121640 SRX10201405 GSM5121641 SRP308746 SRS8347885 GSM5121641 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121641 GSM5121641 GEO Accession GSM5121641 SRX10201406 GSM5121642 SRP308746 SRS8347886 GSM5121642 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121642 GSM5121642 GEO Accession GSM5121642 SRX10201407 GSM5121643 SRP308746 SRS8347887 GSM5121643 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121643 GSM5121643 GEO Accession GSM5121643 SRX10201408 GSM5121644 SRP308746 SRS8347888 GSM5121644 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121644 GSM5121644 GEO Accession GSM5121644 SRX10201409 GSM5121645 SRP308746 SRS8347890 GSM5121645 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121645 GSM5121645 GEO Accession GSM5121645 SRX10201410 GSM5121646 SRP308746 SRS8347889 GSM5121646 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121646 GSM5121646 GEO Accession GSM5121646 SRX10201411 GSM5121647 SRP308746 SRS8347891 GSM5121647 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121647 GSM5121647 GEO Accession GSM5121647 SRX10201412 GSM5121648 SRP308746 SRS8347892 GSM5121648 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121648 GSM5121648 GEO Accession GSM5121648 SRX10201413 GSM5121649 SRP308746 SRS8347897 GSM5121649 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121649 GSM5121649 GEO Accession GSM5121649 SRX10201414 GSM5121650 SRP308746 SRS8347893 GSM5121650 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121650 GSM5121650 GEO Accession GSM5121650 SRX10201415 GSM5121651 SRP308746 SRS8347895 GSM5121651 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121651 GSM5121651 GEO Accession GSM5121651 SRX10201416 GSM5121652 SRP308746 SRS8347894 GSM5121652 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121652 GSM5121652 GEO Accession GSM5121652 SRX10201417 GSM5121653 SRP308746 SRS8347896 GSM5121653 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121653 GSM5121653 GEO Accession GSM5121653 SRX10201418 GSM5121654 SRP308746 SRS8347898 GSM5121654 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121654 GSM5121654 GEO Accession GSM5121654 SRX10201419 GSM5121655 SRP308746 SRS8347903 GSM5121655 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121655 GSM5121655 GEO Accession GSM5121655 SRX10201420 GSM5121656 SRP308746 SRS8347901 GSM5121656 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121656 GSM5121656 GEO Accession GSM5121656 SRX10201421 GSM5121657 SRP308746 SRS8347900 GSM5121657 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121657 GSM5121657 GEO Accession GSM5121657 SRX10201422 GSM5121658 SRP308746 SRS8347899 GSM5121658 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121658 GSM5121658 GEO Accession GSM5121658 SRX10201423 GSM5121659 SRP308746 SRS8347902 GSM5121659 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121659 GSM5121659 GEO Accession GSM5121659 SRX10201424 GSM5121660 SRP308746 SRS8347905 GSM5121660 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121660 GSM5121660 GEO Accession GSM5121660 SRX10201425 GSM5121661 SRP308746 SRS8347904 GSM5121661 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121661 GSM5121661 GEO Accession GSM5121661 SRX10201426 GSM5121662 SRP308746 SRS8347906 GSM5121662 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121662 GSM5121662 GEO Accession GSM5121662 SRX10201427 GSM5121663 SRP308746 SRS8347907 GSM5121663 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121663 GSM5121663 GEO Accession GSM5121663 SRX10201428 GSM5121664 SRP308746 SRS8347908 GSM5121664 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121664 GSM5121664 GEO Accession GSM5121664 SRX10201429 GSM5121665 SRP308746 SRS8347909 GSM5121665 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121665 GSM5121665 GEO Accession GSM5121665 SRX10201430 GSM5121666 SRP308746 SRS8347910 GSM5121666 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121666 GSM5121666 GEO Accession GSM5121666 SRX10201431 GSM5121667 SRP308746 SRS8347911 GSM5121667 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121667 GSM5121667 GEO Accession GSM5121667 SRX10201432 GSM5121668 SRP308746 SRS8347912 GSM5121668 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121668 GSM5121668 GEO Accession GSM5121668 SRX10201433 GSM5121669 SRP308746 SRS8347913 GSM5121669 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121669 GSM5121669 GEO Accession GSM5121669 SRX10201434 GSM5121670 SRP308746 SRS8347914 GSM5121670 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121670 GSM5121670 GEO Accession GSM5121670 SRX10201435 GSM5121671 SRP308746 SRS8347916 GSM5121671 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121671 GSM5121671 GEO Accession GSM5121671 SRX10201436 GSM5121672 SRP308746 SRS8347915 GSM5121672 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121672 GSM5121672 GEO Accession GSM5121672 SRX10201437 GSM5121673 SRP308746 SRS8347918 GSM5121673 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121673 GSM5121673 GEO Accession GSM5121673 SRX10201438 GSM5121674 SRP308746 SRS8347917 GSM5121674 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121674 GSM5121674 GEO Accession GSM5121674 SRX10201439 GSM5121675 SRP308746 SRS8347920 GSM5121675 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121675 GSM5121675 GEO Accession GSM5121675 SRX10201440 GSM5121676 SRP308746 SRS8347919 GSM5121676 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121676 GSM5121676 GEO Accession GSM5121676 SRX10201441 GSM5121677 SRP308746 SRS8347921 GSM5121677 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121677 GSM5121677 GEO Accession GSM5121677 SRX10201442 GSM5121678 SRP308746 SRS8347922 GSM5121678 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121678 GSM5121678 GEO Accession GSM5121678 SRX10201443 GSM5121679 SRP308746 SRS8347923 GSM5121679 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121679 GSM5121679 GEO Accession GSM5121679 SRX10201444 GSM5121680 SRP308746 SRS8347924 GSM5121680 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121680 GSM5121680 GEO Accession GSM5121680 SRX10201445 GSM5121681 SRP308746 SRS8347927 GSM5121681 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121681 GSM5121681 GEO Accession GSM5121681 SRX10201446 GSM5121682 SRP308746 SRS8347928 GSM5121682 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121682 GSM5121682 GEO Accession GSM5121682 SRX10201447 GSM5121683 SRP308746 SRS8347926 GSM5121683 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121683 GSM5121683 GEO Accession GSM5121683 SRX10201448 GSM5121684 SRP308746 SRS8347925 GSM5121684 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121684 GSM5121684 GEO Accession GSM5121684 SRX10201449 GSM5121685 SRP308746 SRS8347929 GSM5121685 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121685 GSM5121685 GEO Accession GSM5121685 SRX10201450 GSM5121686 SRP308746 SRS8347930 GSM5121686 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121686 GSM5121686 GEO Accession GSM5121686 SRX10201451 GSM5121687 SRP308746 SRS8347931 GSM5121687 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121687 GSM5121687 GEO Accession GSM5121687 SRX10201452 GSM5121688 SRP308746 SRS8347932 GSM5121688 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121688 GSM5121688 GEO Accession GSM5121688 SRX10201453 GSM5121689 SRP308746 SRS8347934 GSM5121689 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121689 GSM5121689 GEO Accession GSM5121689 SRX10201454 GSM5121690 SRP308746 SRS8347933 GSM5121690 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121690 GSM5121690 GEO Accession GSM5121690 SRX10201455 GSM5121691 SRP308746 SRS8347935 GSM5121691 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121691 GSM5121691 GEO Accession GSM5121691 SRX10201456 GSM5121692 SRP308746 SRS8347936 GSM5121692 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121692 GSM5121692 GEO Accession GSM5121692 SRX10201457 GSM5121693 SRP308746 SRS8347937 GSM5121693 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121693 GSM5121693 GEO Accession GSM5121693 SRX10201458 GSM5121694 SRP308746 SRS8347938 GSM5121694 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121694 GSM5121694 GEO Accession GSM5121694 SRX10201459 GSM5121695 SRP308746 SRS8347939 GSM5121695 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121695 GSM5121695 GEO Accession GSM5121695 SRX10201460 GSM5121696 SRP308746 SRS8347940 GSM5121696 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121696 GSM5121696 GEO Accession GSM5121696 SRX10201461 GSM5121697 SRP308746 SRS8347941 GSM5121697 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121697 GSM5121697 GEO Accession GSM5121697 SRX10201462 GSM5121698 SRP308746 SRS8347942 GSM5121698 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121698 GSM5121698 GEO Accession GSM5121698 SRX10201463 GSM5121699 SRP308746 SRS8347943 GSM5121699 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121699 GSM5121699 GEO Accession GSM5121699 SRX10201464 GSM5121700 SRP308746 SRS8347944 GSM5121700 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121700 GSM5121700 GEO Accession GSM5121700 SRX10201465 GSM5121701 SRP308746 SRS8347945 GSM5121701 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121701 GSM5121701 GEO Accession GSM5121701 SRX10201466 GSM5121702 SRP308746 SRS8347946 GSM5121702 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121702 GSM5121702 GEO Accession GSM5121702 SRX10201467 GSM5121703 SRP308746 SRS8347947 GSM5121703 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121703 GSM5121703 GEO Accession GSM5121703 SRX10201468 GSM5121704 SRP308746 SRS8347949 GSM5121704 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121704 GSM5121704 GEO Accession GSM5121704 SRX10201469 GSM5121705 SRP308746 SRS8347948 GSM5121705 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121705 GSM5121705 GEO Accession GSM5121705 SRX10201470 GSM5121706 SRP308746 SRS8347950 GSM5121706 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121706 GSM5121706 GEO Accession GSM5121706 SRX10201471 GSM5121707 SRP308746 SRS8347954 GSM5121707 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121707 GSM5121707 GEO Accession GSM5121707 SRX10201472 GSM5121708 SRP308746 SRS8347951 GSM5121708 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121708 GSM5121708 GEO Accession GSM5121708 SRX10201473 GSM5121709 SRP308746 SRS8347952 GSM5121709 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121709 GSM5121709 GEO Accession GSM5121709 SRX10201474 GSM5121710 SRP308746 SRS8347953 GSM5121710 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121710 GSM5121710 GEO Accession GSM5121710 SRX10201475 GSM5121711 SRP308746 SRS8347956 GSM5121711 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121711 GSM5121711 GEO Accession GSM5121711 SRX10201476 GSM5121712 SRP308746 SRS8347955 GSM5121712 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121712 GSM5121712 GEO Accession GSM5121712 SRX10201477 GSM5121713 SRP308746 SRS8347958 GSM5121713 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121713 GSM5121713 GEO Accession GSM5121713 SRX10201478 GSM5121714 SRP308746 SRS8347957 GSM5121714 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121714 GSM5121714 GEO Accession GSM5121714 SRX10201479 GSM5121715 SRP308746 SRS8347959 GSM5121715 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121715 GSM5121715 GEO Accession GSM5121715 SRX10201480 GSM5121716 SRP308746 SRS8347960 GSM5121716 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121716 GSM5121716 GEO Accession GSM5121716 SRX10201481 GSM5121717 SRP308746 SRS8347961 GSM5121717 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121717 GSM5121717 GEO Accession GSM5121717 SRX10201482 GSM5121718 SRP308746 SRS8347962 GSM5121718 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121718 GSM5121718 GEO Accession GSM5121718 SRX10201483 GSM5121719 SRP308746 SRS8347963 GSM5121719 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121719 GSM5121719 GEO Accession GSM5121719 SRX10201484 GSM5121720 SRP308746 SRS8347964 GSM5121720 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121720 GSM5121720 GEO Accession GSM5121720 SRX10201485 GSM5121721 SRP308746 SRS8347965 GSM5121721 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121721 GSM5121721 GEO Accession GSM5121721 SRX10201486 GSM5121722 SRP308746 SRS8347966 GSM5121722 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121722 GSM5121722 GEO Accession GSM5121722 SRX10201487 GSM5121723 SRP308746 SRS8347967 GSM5121723 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121723 GSM5121723 GEO Accession GSM5121723 SRX10201488 GSM5121724 SRP308746 SRS8347969 GSM5121724 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121724 GSM5121724 GEO Accession GSM5121724 SRX10201489 GSM5121725 SRP308746 SRS8347968 GSM5121725 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121725 GSM5121725 GEO Accession GSM5121725 SRX10201490 GSM5121726 SRP308746 SRS8347970 GSM5121726 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121726 GSM5121726 GEO Accession GSM5121726 SRX10201491 GSM5121727 SRP308746 SRS8347971 GSM5121727 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121727 GSM5121727 GEO Accession GSM5121727 SRX10201492 GSM5121728 SRP308746 SRS8347972 GSM5121728 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121728 GSM5121728 GEO Accession GSM5121728 SRX10201493 GSM5121729 SRP308746 SRS8347975 GSM5121729 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121729 GSM5121729 GEO Accession GSM5121729 SRX10201494 GSM5121730 SRP308746 SRS8347973 GSM5121730 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121730 GSM5121730 GEO Accession GSM5121730 SRX10201495 GSM5121731 SRP308746 SRS8347974 GSM5121731 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121731 GSM5121731 GEO Accession GSM5121731 SRX10201496 GSM5121732 SRP308746 SRS8347976 GSM5121732 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121732 GSM5121732 GEO Accession GSM5121732 SRX10201497 GSM5121733 SRP308746 SRS8347977 GSM5121733 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121733 GSM5121733 GEO Accession GSM5121733 SRX10201498 GSM5121734 SRP308746 SRS8347979 GSM5121734 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121734 GSM5121734 GEO Accession GSM5121734 SRX10201499 GSM5121735 SRP308746 SRS8347978 GSM5121735 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121735 GSM5121735 GEO Accession GSM5121735 SRX10201500 GSM5121736 SRP308746 SRS8347980 GSM5121736 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121736 GSM5121736 GEO Accession GSM5121736 SRX10201501 GSM5121737 SRP308746 SRS8347981 GSM5121737 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121737 GSM5121737 GEO Accession GSM5121737 SRX10201502 GSM5121738 SRP308746 SRS8347982 GSM5121738 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121738 GSM5121738 GEO Accession GSM5121738 SRX10201503 GSM5121739 SRP308746 SRS8347984 GSM5121739 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121739 GSM5121739 GEO Accession GSM5121739 SRX10201504 GSM5121740 SRP308746 SRS8347983 GSM5121740 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121740 GSM5121740 GEO Accession GSM5121740 SRX10201505 GSM5121741 SRP308746 SRS8347985 GSM5121741 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121741 GSM5121741 GEO Accession GSM5121741 SRX10201506 GSM5121742 SRP308746 SRS8347986 GSM5121742 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121742 GSM5121742 GEO Accession GSM5121742 SRX10201507 GSM5121743 SRP308746 SRS8347987 GSM5121743 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121743 GSM5121743 GEO Accession GSM5121743 SRX10201508 GSM5121744 SRP308746 SRS8347988 GSM5121744 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121744 GSM5121744 GEO Accession GSM5121744 SRX10201509 GSM5121745 SRP308746 SRS8347989 GSM5121745 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121745 GSM5121745 GEO Accession GSM5121745 SRX10201510 GSM5121746 SRP308746 SRS8347990 GSM5121746 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121746 GSM5121746 GEO Accession GSM5121746 SRX10201511 GSM5121747 SRP308746 SRS8347991 GSM5121747 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121747 GSM5121747 GEO Accession GSM5121747 SRX10201512 GSM5121748 SRP308746 SRS8347992 GSM5121748 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121748 GSM5121748 GEO Accession GSM5121748 SRX10201513 GSM5121749 SRP308746 SRS8347993 GSM5121749 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121749 GSM5121749 GEO Accession GSM5121749 SRX10201514 GSM5121750 SRP308746 SRS8347994 GSM5121750 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121750 GSM5121750 GEO Accession GSM5121750 SRX10201515 GSM5121751 SRP308746 SRS8347995 GSM5121751 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121751 GSM5121751 GEO Accession GSM5121751 SRX10201516 GSM5121752 SRP308746 SRS8347996 GSM5121752 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121752 GSM5121752 GEO Accession GSM5121752 SRX10201517 GSM5121753 SRP308746 SRS8347997 GSM5121753 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121753 GSM5121753 GEO Accession GSM5121753 SRX10201518 GSM5121754 SRP308746 SRS8347998 GSM5121754 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121754 GSM5121754 GEO Accession GSM5121754 SRX10201519 GSM5121755 SRP308746 SRS8347999 GSM5121755 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121755 GSM5121755 GEO Accession GSM5121755 SRX10201520 GSM5121756 SRP308746 SRS8348000 GSM5121756 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121756 GSM5121756 GEO Accession GSM5121756 SRX10201521 GSM5121757 SRP308746 SRS8348001 GSM5121757 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121757 GSM5121757 GEO Accession GSM5121757 SRX10201522 GSM5121758 SRP308746 SRS8348003 GSM5121758 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121758 GSM5121758 GEO Accession GSM5121758 SRX10201523 GSM5121759 SRP308746 SRS8348002 GSM5121759 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121759 GSM5121759 GEO Accession GSM5121759 SRX10201524 GSM5121760 SRP308746 SRS8348004 GSM5121760 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121760 GSM5121760 GEO Accession GSM5121760 SRX10201525 GSM5121761 SRP308746 SRS8348006 GSM5121761 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121761 GSM5121761 GEO Accession GSM5121761 SRX10201526 GSM5121762 SRP308746 SRS8348005 GSM5121762 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121762 GSM5121762 GEO Accession GSM5121762 SRX10201527 GSM5121763 SRP308746 SRS8348007 GSM5121763 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121763 GSM5121763 GEO Accession GSM5121763 SRX10201528 GSM5121764 SRP308746 SRS8348008 GSM5121764 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121764 GSM5121764 GEO Accession GSM5121764 SRX10201529 GSM5121765 SRP308746 SRS8348009 GSM5121765 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121765 GSM5121765 GEO Accession GSM5121765 SRX10201530 GSM5121766 SRP308746 SRS8348010 GSM5121766 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121766 GSM5121766 GEO Accession GSM5121766 SRX10201531 GSM5121767 SRP308746 SRS8348011 GSM5121767 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121767 GSM5121767 GEO Accession GSM5121767 SRX10201532 GSM5121768 SRP308746 SRS8348012 GSM5121768 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121768 GSM5121768 GEO Accession GSM5121768 SRX10201533 GSM5121769 SRP308746 SRS8348013 GSM5121769 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121769 GSM5121769 GEO Accession GSM5121769 SRX10201534 GSM5121770 SRP308746 SRS8348014 GSM5121770 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121770 GSM5121770 GEO Accession GSM5121770 SRX10201535 GSM5121771 SRP308746 SRS8348015 GSM5121771 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121771 GSM5121771 GEO Accession GSM5121771 SRX10201536 GSM5121772 SRP308746 SRS8348016 GSM5121772 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121772 GSM5121772 GEO Accession GSM5121772 SRX10201537 GSM5121773 SRP308746 SRS8348017 GSM5121773 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121773 GSM5121773 GEO Accession GSM5121773 SRX10201538 GSM5121774 SRP308746 SRS8348018 GSM5121774 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121774 GSM5121774 GEO Accession GSM5121774 SRX10201539 GSM5121775 SRP308746 SRS8348020 GSM5121775 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121775 GSM5121775 GEO Accession GSM5121775 SRX10201540 GSM5121776 SRP308746 SRS8348019 GSM5121776 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121776 GSM5121776 GEO Accession GSM5121776 SRX10201541 GSM5121777 SRP308746 SRS8348021 GSM5121777 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121777 GSM5121777 GEO Accession GSM5121777 SRX10201542 GSM5121778 SRP308746 SRS8348022 GSM5121778 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121778 GSM5121778 GEO Accession GSM5121778 SRX10201543 GSM5121779 SRP308746 SRS8348023 GSM5121779 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121779 GSM5121779 GEO Accession GSM5121779 SRX10201544 GSM5121780 SRP308746 SRS8348024 GSM5121780 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121780 GSM5121780 GEO Accession GSM5121780 SRX10201545 GSM5121781 SRP308746 SRS8348026 GSM5121781 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121781 GSM5121781 GEO Accession GSM5121781 SRX10201546 GSM5121782 SRP308746 SRS8348025 GSM5121782 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121782 GSM5121782 GEO Accession GSM5121782 SRX10201547 GSM5121783 SRP308746 SRS8348027 GSM5121783 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121783 GSM5121783 GEO Accession GSM5121783 SRX10201548 GSM5121784 SRP308746 SRS8348028 GSM5121784 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121784 GSM5121784 GEO Accession GSM5121784 SRX10201549 GSM5121785 SRP308746 SRS8348029 GSM5121785 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121785 GSM5121785 GEO Accession GSM5121785 SRX10201550 GSM5121786 SRP308746 SRS8348031 GSM5121786 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121786 GSM5121786 GEO Accession GSM5121786 SRX10201551 GSM5121787 SRP308746 SRS8348030 GSM5121787 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121787 GSM5121787 GEO Accession GSM5121787 SRX10201552 GSM5121788 SRP308746 SRS8348033 GSM5121788 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121788 GSM5121788 GEO Accession GSM5121788 SRX10201553 GSM5121789 SRP308746 SRS8348032 GSM5121789 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121789 GSM5121789 GEO Accession GSM5121789 SRX10201554 GSM5121790 SRP308746 SRS8348034 GSM5121790 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121790 GSM5121790 GEO Accession GSM5121790 SRX10201555 GSM5121791 SRP308746 SRS8348035 GSM5121791 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121791 GSM5121791 GEO Accession GSM5121791 SRX10201556 GSM5121792 SRP308746 SRS8348036 GSM5121792 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121792 GSM5121792 GEO Accession GSM5121792 SRX10201557 GSM5121793 SRP308746 SRS8348037 GSM5121793 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121793 GSM5121793 GEO Accession GSM5121793 SRX10201558 GSM5121794 SRP308746 SRS8348038 GSM5121794 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121794 GSM5121794 GEO Accession GSM5121794 SRX10201559 GSM5121795 SRP308746 SRS8348039 GSM5121795 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121795 GSM5121795 GEO Accession GSM5121795 SRX10201560 GSM5121796 SRP308746 SRS8348040 GSM5121796 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121796 GSM5121796 GEO Accession GSM5121796 SRX10201561 GSM5121797 SRP308746 SRS8348041 GSM5121797 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121797 GSM5121797 GEO Accession GSM5121797 SRX10201562 GSM5121798 SRP308746 SRS8348044 GSM5121798 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121798 GSM5121798 GEO Accession GSM5121798 SRX10201563 GSM5121799 SRP308746 SRS8348043 GSM5121799 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121799 GSM5121799 GEO Accession GSM5121799 SRX10201564 GSM5121800 SRP308746 SRS8348042 GSM5121800 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121800 GSM5121800 GEO Accession GSM5121800 SRX10201565 GSM5121801 SRP308746 SRS8348047 GSM5121801 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121801 GSM5121801 GEO Accession GSM5121801 SRX10201566 GSM5121802 SRP308746 SRS8348045 GSM5121802 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121802 GSM5121802 GEO Accession GSM5121802 SRX10201567 GSM5121803 SRP308746 SRS8348046 GSM5121803 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121803 GSM5121803 GEO Accession GSM5121803 SRX10201568 GSM5121804 SRP308746 SRS8348048 GSM5121804 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121804 GSM5121804 GEO Accession GSM5121804 SRX10201569 GSM5121805 SRP308746 SRS8348049 GSM5121805 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121805 GSM5121805 GEO Accession GSM5121805 SRX10201570 GSM5121806 SRP308746 SRS8348050 GSM5121806 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121806 GSM5121806 GEO Accession GSM5121806 SRX10201571 GSM5121807 SRP308746 SRS8348053 GSM5121807 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121807 GSM5121807 GEO Accession GSM5121807 SRX10201572 GSM5121808 SRP308746 SRS8348051 GSM5121808 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121808 GSM5121808 GEO Accession GSM5121808 SRX10201573 GSM5121809 SRP308746 SRS8348052 GSM5121809 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121809 GSM5121809 GEO Accession GSM5121809 SRX10201574 GSM5121810 SRP308746 SRS8348054 GSM5121810 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121810 GSM5121810 GEO Accession GSM5121810 SRX10201575 GSM5121811 SRP308746 SRS8348055 GSM5121811 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121811 GSM5121811 GEO Accession GSM5121811 SRX10201576 GSM5121812 SRP308746 SRS8348056 GSM5121812 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121812 GSM5121812 GEO Accession GSM5121812 SRX10201577 GSM5121813 SRP308746 SRS8348057 GSM5121813 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121813 GSM5121813 GEO Accession GSM5121813 SRX10201578 GSM5121814 SRP308746 SRS8348059 GSM5121814 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121814 GSM5121814 GEO Accession GSM5121814 SRX10201579 GSM5121815 SRP308746 SRS8348058 GSM5121815 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121815 GSM5121815 GEO Accession GSM5121815 SRX10201580 GSM5121816 SRP308746 SRS8348060 GSM5121816 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121816 GSM5121816 GEO Accession GSM5121816 SRX10201581 GSM5121817 SRP308746 SRS8348062 GSM5121817 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121817 GSM5121817 GEO Accession GSM5121817 SRX10201582 GSM5121818 SRP308746 SRS8348061 GSM5121818 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121818 GSM5121818 GEO Accession GSM5121818 SRX10201583 GSM5121819 SRP308746 SRS8348063 GSM5121819 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121819 GSM5121819 GEO Accession GSM5121819 SRX10201584 GSM5121820 SRP308746 SRS8348065 GSM5121820 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121820 GSM5121820 GEO Accession GSM5121820 SRX10201585 GSM5121821 SRP308746 SRS8348066 GSM5121821 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121821 GSM5121821 GEO Accession GSM5121821 SRX10201586 GSM5121822 SRP308746 SRS8348064 GSM5121822 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121822 GSM5121822 GEO Accession GSM5121822 SRX10201587 GSM5121823 SRP308746 SRS8348067 GSM5121823 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121823 GSM5121823 GEO Accession GSM5121823 SRX10201588 GSM5121824 SRP308746 SRS8348068 GSM5121824 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121824 GSM5121824 GEO Accession GSM5121824 SRX10201589 GSM5121825 SRP308746 SRS8348069 GSM5121825 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121825 GSM5121825 GEO Accession GSM5121825 SRX10201590 GSM5121826 SRP308746 SRS8348070 GSM5121826 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121826 GSM5121826 GEO Accession GSM5121826 SRX10201591 GSM5121827 SRP308746 SRS8348072 GSM5121827 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121827 GSM5121827 GEO Accession GSM5121827 SRX10201592 GSM5121828 SRP308746 SRS8348071 GSM5121828 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121828 GSM5121828 GEO Accession GSM5121828 SRX10201593 GSM5121829 SRP308746 SRS8348075 GSM5121829 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121829 GSM5121829 GEO Accession GSM5121829 SRX10201594 GSM5121830 SRP308746 SRS8348073 GSM5121830 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121830 GSM5121830 GEO Accession GSM5121830 SRX10201595 GSM5121831 SRP308746 SRS8348074 GSM5121831 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121831 GSM5121831 GEO Accession GSM5121831 SRX10201596 GSM5121832 SRP308746 SRS8348076 GSM5121832 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121832 GSM5121832 GEO Accession GSM5121832 SRX10201597 GSM5121833 SRP308746 SRS8348077 GSM5121833 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121833 GSM5121833 GEO Accession GSM5121833 SRX10201598 GSM5121834 SRP308746 SRS8348078 GSM5121834 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121834 GSM5121834 GEO Accession GSM5121834 SRX10201599 GSM5121835 SRP308746 SRS8348081 GSM5121835 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121835 GSM5121835 GEO Accession GSM5121835 SRX10201600 GSM5121836 SRP308746 SRS8348080 GSM5121836 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121836 GSM5121836 GEO Accession GSM5121836 SRX10201601 GSM5121837 SRP308746 SRS8348079 GSM5121837 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121837 GSM5121837 GEO Accession GSM5121837 SRX10201602 GSM5121838 SRP308746 SRS8348082 GSM5121838 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121838 GSM5121838 GEO Accession GSM5121838 SRX10201603 GSM5121839 SRP308746 SRS8348083 GSM5121839 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121839 GSM5121839 GEO Accession GSM5121839 SRX10201604 GSM5121840 SRP308746 SRS8348085 GSM5121840 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121840 GSM5121840 GEO Accession GSM5121840 SRX10201605 GSM5121841 SRP308746 SRS8348086 GSM5121841 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121841 GSM5121841 GEO Accession GSM5121841 SRX10201606 GSM5121842 SRP308746 SRS8348084 GSM5121842 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121842 GSM5121842 GEO Accession GSM5121842 SRX10201607 GSM5121843 SRP308746 SRS8348088 GSM5121843 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121843 GSM5121843 GEO Accession GSM5121843 SRX10201608 GSM5121844 SRP308746 SRS8348087 GSM5121844 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121844 GSM5121844 GEO Accession GSM5121844 SRX10201609 GSM5121845 SRP308746 SRS8348089 GSM5121845 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121845 GSM5121845 GEO Accession GSM5121845 SRX10201610 GSM5121846 SRP308746 SRS8348091 GSM5121846 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121846 GSM5121846 GEO Accession GSM5121846 SRX10201611 GSM5121847 SRP308746 SRS8348090 GSM5121847 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121847 GSM5121847 GEO Accession GSM5121847 SRX10201612 GSM5121848 SRP308746 SRS8348092 GSM5121848 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121848 GSM5121848 GEO Accession GSM5121848 SRX10201613 GSM5121849 SRP308746 SRS8348093 GSM5121849 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121849 GSM5121849 GEO Accession GSM5121849 SRX10201614 GSM5121850 SRP308746 SRS8348095 GSM5121850 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121850 GSM5121850 GEO Accession GSM5121850 SRX10201615 GSM5121851 SRP308746 SRS8348094 GSM5121851 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121851 GSM5121851 GEO Accession GSM5121851 SRX10201616 GSM5121852 SRP308746 SRS8348096 GSM5121852 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121852 GSM5121852 GEO Accession GSM5121852 SRX10201617 GSM5121853 SRP308746 SRS8348098 GSM5121853 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121853 GSM5121853 GEO Accession GSM5121853 SRX10201618 GSM5121854 SRP308746 SRS8348097 GSM5121854 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121854 GSM5121854 GEO Accession GSM5121854 SRX10201619 GSM5121855 SRP308746 SRS8348100 GSM5121855 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121855 GSM5121855 GEO Accession GSM5121855 SRX10201620 GSM5121856 SRP308746 SRS8348101 GSM5121856 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121856 GSM5121856 GEO Accession GSM5121856 SRX10201621 GSM5121857 SRP308746 SRS8348099 GSM5121857 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121857 GSM5121857 GEO Accession GSM5121857 SRX10201622 GSM5121858 SRP308746 SRS8348102 GSM5121858 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121858 GSM5121858 GEO Accession GSM5121858 SRX10201623 GSM5121859 SRP308746 SRS8348104 GSM5121859 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121859 GSM5121859 GEO Accession GSM5121859 SRX10201624 GSM5121860 SRP308746 SRS8348103 GSM5121860 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121860 GSM5121860 GEO Accession GSM5121860 SRX10201625 GSM5121861 SRP308746 SRS8348105 GSM5121861 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121861 GSM5121861 GEO Accession GSM5121861 SRX10201626 GSM5121862 SRP308746 SRS8348107 GSM5121862 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121862 GSM5121862 GEO Accession GSM5121862 SRX10201627 GSM5121863 SRP308746 SRS8348108 GSM5121863 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121863 GSM5121863 GEO Accession GSM5121863 SRX10201628 GSM5121864 SRP308746 SRS8348106 GSM5121864 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121864 GSM5121864 GEO Accession GSM5121864 SRX10201629 GSM5121865 SRP308746 SRS8348109 GSM5121865 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121865 GSM5121865 GEO Accession GSM5121865 SRX10201630 GSM5121866 SRP308746 SRS8348110 GSM5121866 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121866 GSM5121866 GEO Accession GSM5121866 SRX10201631 GSM5121867 SRP308746 SRS8348112 GSM5121867 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121867 GSM5121867 GEO Accession GSM5121867 SRX10201632 GSM5121868 SRP308746 SRS8348111 GSM5121868 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121868 GSM5121868 GEO Accession GSM5121868 SRX10201633 GSM5121869 SRP308746 SRS8348114 GSM5121869 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121869 GSM5121869 GEO Accession GSM5121869 SRX10201634 GSM5121870 SRP308746 SRS8348115 GSM5121870 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121870 GSM5121870 GEO Accession GSM5121870 SRX10201635 GSM5121871 SRP308746 SRS8348113 GSM5121871 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121871 GSM5121871 GEO Accession GSM5121871 SRX10201636 GSM5121872 SRP308746 SRS8348116 GSM5121872 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121872 GSM5121872 GEO Accession GSM5121872 SRX10201637 GSM5121873 SRP308746 SRS8348118 GSM5121873 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121873 GSM5121873 GEO Accession GSM5121873 SRX10201638 GSM5121874 SRP308746 SRS8348117 GSM5121874 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121874 GSM5121874 GEO Accession GSM5121874 SRX10201639 GSM5121875 SRP308746 SRS8348120 GSM5121875 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121875 GSM5121875 GEO Accession GSM5121875 SRX10201640 GSM5121876 SRP308746 SRS8348119 GSM5121876 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121876 GSM5121876 GEO Accession GSM5121876 SRX10201641 GSM5121877 SRP308746 SRS8348122 GSM5121877 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121877 GSM5121877 GEO Accession GSM5121877 SRX10201642 GSM5121878 SRP308746 SRS8348121 GSM5121878 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121878 GSM5121878 GEO Accession GSM5121878 SRX10201643 GSM5121879 SRP308746 SRS8348123 GSM5121879 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121879 GSM5121879 GEO Accession GSM5121879 SRX10201644 GSM5121880 SRP308746 SRS8348124 GSM5121880 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121880 GSM5121880 GEO Accession GSM5121880 SRX10201645 GSM5121881 SRP308746 SRS8348125 GSM5121881 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121881 GSM5121881 GEO Accession GSM5121881 SRX10201646 GSM5121882 SRP308746 SRS8348126 GSM5121882 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121882 GSM5121882 GEO Accession GSM5121882 SRX10201647 GSM5121883 SRP308746 SRS8348129 GSM5121883 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121883 GSM5121883 GEO Accession GSM5121883 SRX10201648 GSM5121884 SRP308746 SRS8348130 GSM5121884 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121884 GSM5121884 GEO Accession GSM5121884 SRX10201649 GSM5121885 SRP308746 SRS8348127 GSM5121885 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121885 GSM5121885 GEO Accession GSM5121885 SRX10201650 GSM5121886 SRP308746 SRS8348128 GSM5121886 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121886 GSM5121886 GEO Accession GSM5121886 SRX10201651 GSM5121887 SRP308746 SRS8348132 GSM5121887 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121887 GSM5121887 GEO Accession GSM5121887 SRX10201652 GSM5121888 SRP308746 SRS8348134 GSM5121888 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121888 GSM5121888 GEO Accession GSM5121888 SRX10201653 GSM5121889 SRP308746 SRS8348133 GSM5121889 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121889 GSM5121889 GEO Accession GSM5121889 SRX10201654 GSM5121890 SRP308746 SRS8348131 GSM5121890 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121890 GSM5121890 GEO Accession GSM5121890 SRX10201655 GSM5121891 SRP308746 SRS8348135 GSM5121891 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121891 GSM5121891 GEO Accession GSM5121891 SRX10201656 GSM5121892 SRP308746 SRS8348138 GSM5121892 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121892 GSM5121892 GEO Accession GSM5121892 SRX10201657 GSM5121893 SRP308746 SRS8348136 GSM5121893 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121893 GSM5121893 GEO Accession GSM5121893 SRX10201658 GSM5121894 SRP308746 SRS8348137 GSM5121894 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121894 GSM5121894 GEO Accession GSM5121894 SRX10201659 GSM5121895 SRP308746 SRS8348139 GSM5121895 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121895 GSM5121895 GEO Accession GSM5121895 SRX10201660 GSM5121896 SRP308746 SRS8348140 GSM5121896 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121896 GSM5121896 GEO Accession GSM5121896 SRX10201661 GSM5121897 SRP308746 SRS8348141 GSM5121897 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121897 GSM5121897 GEO Accession GSM5121897 SRX10201662 GSM5121898 SRP308746 SRS8348143 GSM5121898 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121898 GSM5121898 GEO Accession GSM5121898 SRX10201663 GSM5121899 SRP308746 SRS8348145 GSM5121899 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121899 GSM5121899 GEO Accession GSM5121899 SRX10201664 GSM5121900 SRP308746 SRS8348142 GSM5121900 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121900 GSM5121900 GEO Accession GSM5121900 SRX10201665 GSM5121901 SRP308746 SRS8348144 GSM5121901 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121901 GSM5121901 GEO Accession GSM5121901 SRX10201666 GSM5121902 SRP308746 SRS8348146 GSM5121902 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121902 GSM5121902 GEO Accession GSM5121902 SRX10201667 GSM5121903 SRP308746 SRS8348147 GSM5121903 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121903 GSM5121903 GEO Accession GSM5121903 SRX10201668 GSM5121904 SRP308746 SRS8348149 GSM5121904 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121904 GSM5121904 GEO Accession GSM5121904 SRX10201669 GSM5121905 SRP308746 SRS8348148 GSM5121905 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121905 GSM5121905 GEO Accession GSM5121905 SRX10201670 GSM5121906 SRP308746 SRS8348151 GSM5121906 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121906 GSM5121906 GEO Accession GSM5121906 SRX10201671 GSM5121907 SRP308746 SRS8348150 GSM5121907 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121907 GSM5121907 GEO Accession GSM5121907 SRX10201672 GSM5121908 SRP308746 SRS8348153 GSM5121908 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121908 GSM5121908 GEO Accession GSM5121908 SRX10201673 GSM5121909 SRP308746 SRS8348152 GSM5121909 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121909 GSM5121909 GEO Accession GSM5121909 SRX10201674 GSM5121910 SRP308746 SRS8348154 GSM5121910 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121910 GSM5121910 GEO Accession GSM5121910 SRX10201675 GSM5121911 SRP308746 SRS8348155 GSM5121911 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121911 GSM5121911 GEO Accession GSM5121911 SRX10201676 GSM5121912 SRP308746 SRS8348156 GSM5121912 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121912 GSM5121912 GEO Accession GSM5121912 SRX10201677 GSM5121913 SRP308746 SRS8348157 GSM5121913 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121913 GSM5121913 GEO Accession GSM5121913 SRX10201678 GSM5121914 SRP308746 SRS8348160 GSM5121914 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121914 GSM5121914 GEO Accession GSM5121914 SRX10201679 GSM5121915 SRP308746 SRS8348158 GSM5121915 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121915 GSM5121915 GEO Accession GSM5121915 SRX10201680 GSM5121916 SRP308746 SRS8348159 GSM5121916 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121916 GSM5121916 GEO Accession GSM5121916 SRX10201681 GSM5121917 SRP308746 SRS8348161 GSM5121917 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121917 GSM5121917 GEO Accession GSM5121917 SRX10201682 GSM5121918 SRP308746 SRS8348164 GSM5121918 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121918 GSM5121918 GEO Accession GSM5121918 SRX10201683 GSM5121919 SRP308746 SRS8348162 GSM5121919 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121919 GSM5121919 GEO Accession GSM5121919 SRX10201684 GSM5121920 SRP308746 SRS8348163 GSM5121920 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121920 GSM5121920 GEO Accession GSM5121920 SRX10201685 GSM5121921 SRP308746 SRS8348165 GSM5121921 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121921 GSM5121921 GEO Accession GSM5121921 SRX10201686 GSM5121922 SRP308746 SRS8348166 GSM5121922 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121922 GSM5121922 GEO Accession GSM5121922 SRX10201687 GSM5121923 SRP308746 SRS8348168 GSM5121923 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121923 GSM5121923 GEO Accession GSM5121923 SRX10201688 GSM5121924 SRP308746 SRS8348167 GSM5121924 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121924 GSM5121924 GEO Accession GSM5121924 SRX10201689 GSM5121925 SRP308746 SRS8348169 GSM5121925 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121925 GSM5121925 GEO Accession GSM5121925 SRX10201690 GSM5121926 SRP308746 SRS8348170 GSM5121926 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121926 GSM5121926 GEO Accession GSM5121926 SRX10201691 GSM5121927 SRP308746 SRS8348171 GSM5121927 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121927 GSM5121927 GEO Accession GSM5121927 SRX10201692 GSM5121928 SRP308746 SRS8348172 GSM5121928 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121928 GSM5121928 GEO Accession GSM5121928 SRX10201693 GSM5121929 SRP308746 SRS8348174 GSM5121929 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121929 GSM5121929 GEO Accession GSM5121929 SRX10201694 GSM5121930 SRP308746 SRS8348173 GSM5121930 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121930 GSM5121930 GEO Accession GSM5121930 SRX10201695 GSM5121931 SRP308746 SRS8348175 GSM5121931 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121931 GSM5121931 GEO Accession GSM5121931 SRX10201696 GSM5121932 SRP308746 SRS8348176 GSM5121932 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121932 GSM5121932 GEO Accession GSM5121932 SRX10201697 GSM5121933 SRP308746 SRS8348177 GSM5121933 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121933 GSM5121933 GEO Accession GSM5121933 SRX10201698 GSM5121934 SRP308746 SRS8348178 GSM5121934 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121934 GSM5121934 GEO Accession GSM5121934 SRX10201699 GSM5121935 SRP308746 SRS8348179 GSM5121935 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121935 GSM5121935 GEO Accession GSM5121935 SRX10201700 GSM5121936 SRP308746 SRS8348180 GSM5121936 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121936 GSM5121936 GEO Accession GSM5121936 SRX10201701 GSM5121937 SRP308746 SRS8348182 GSM5121937 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121937 GSM5121937 GEO Accession GSM5121937 SRX10201702 GSM5121938 SRP308746 SRS8348181 GSM5121938 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121938 GSM5121938 GEO Accession GSM5121938 SRX10201703 GSM5121939 SRP308746 SRS8348184 GSM5121939 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121939 GSM5121939 GEO Accession GSM5121939 SRX10201704 GSM5121940 SRP308746 SRS8348183 GSM5121940 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121940 GSM5121940 GEO Accession GSM5121940 SRX10201705 GSM5121941 SRP308746 SRS8348185 GSM5121941 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121941 GSM5121941 GEO Accession GSM5121941 SRX10201706 GSM5121942 SRP308746 SRS8348186 GSM5121942 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121942 GSM5121942 GEO Accession GSM5121942 SRX10201707 GSM5121943 SRP308746 SRS8348187 GSM5121943 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121943 GSM5121943 GEO Accession GSM5121943 SRX10201708 GSM5121944 SRP308746 SRS8348188 GSM5121944 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121944 GSM5121944 GEO Accession GSM5121944 SRX10201709 GSM5121945 SRP308746 SRS8348189 GSM5121945 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121945 GSM5121945 GEO Accession GSM5121945 SRX10201710 GSM5121946 SRP308746 SRS8348190 GSM5121946 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121946 GSM5121946 GEO Accession GSM5121946 SRX10201711 GSM5121947 SRP308746 SRS8348192 GSM5121947 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121947 GSM5121947 GEO Accession GSM5121947 SRX10201712 GSM5121948 SRP308746 SRS8348191 GSM5121948 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121948 GSM5121948 GEO Accession GSM5121948 SRX10201713 GSM5121949 SRP308746 SRS8348194 GSM5121949 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121949 GSM5121949 GEO Accession GSM5121949 SRX10201714 GSM5121950 SRP308746 SRS8348193 GSM5121950 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121950 GSM5121950 GEO Accession GSM5121950 SRX10201715 GSM5121951 SRP308746 SRS8348195 GSM5121951 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121951 GSM5121951 GEO Accession GSM5121951 SRX10201716 GSM5121952 SRP308746 SRS8348196 GSM5121952 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121952 GSM5121952 GEO Accession GSM5121952 SRX10201717 GSM5121953 SRP308746 SRS8348199 GSM5121953 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121953 GSM5121953 GEO Accession GSM5121953 SRX10201718 GSM5121954 SRP308746 SRS8348197 GSM5121954 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121954 GSM5121954 GEO Accession GSM5121954 SRX10201719 GSM5121955 SRP308746 SRS8348198 GSM5121955 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121955 GSM5121955 GEO Accession GSM5121955 SRX10201720 GSM5121956 SRP308746 SRS8348201 GSM5121956 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121956 GSM5121956 GEO Accession GSM5121956 SRX10201721 GSM5121957 SRP308746 SRS8348200 GSM5121957 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121957 GSM5121957 GEO Accession GSM5121957 SRX10201722 GSM5121958 SRP308746 SRS8348203 GSM5121958 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121958 GSM5121958 GEO Accession GSM5121958 SRX10201723 GSM5121959 SRP308746 SRS8348202 GSM5121959 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121959 GSM5121959 GEO Accession GSM5121959 SRX10201724 GSM5121960 SRP308746 SRS8348205 GSM5121960 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121960 GSM5121960 GEO Accession GSM5121960 SRX10201725 GSM5121961 SRP308746 SRS8348204 GSM5121961 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121961 GSM5121961 GEO Accession GSM5121961 SRX10201726 GSM5121962 SRP308746 SRS8348206 GSM5121962 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121962 GSM5121962 GEO Accession GSM5121962 SRX10201727 GSM5121963 SRP308746 SRS8348207 GSM5121963 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121963 GSM5121963 GEO Accession GSM5121963 SRX10201728 GSM5121964 SRP308746 SRS8348209 GSM5121964 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121964 GSM5121964 GEO Accession GSM5121964 SRX10201729 GSM5121965 SRP308746 SRS8348208 GSM5121965 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121965 GSM5121965 GEO Accession GSM5121965 SRX10201730 GSM5121966 SRP308746 SRS8348210 GSM5121966 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121966 GSM5121966 GEO Accession GSM5121966 SRX10201731 GSM5121967 SRP308746 SRS8348213 GSM5121967 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121967 GSM5121967 GEO Accession GSM5121967 SRX10201732 GSM5121968 SRP308746 SRS8348211 GSM5121968 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121968 GSM5121968 GEO Accession GSM5121968 SRX10201733 GSM5121969 SRP308746 SRS8348212 GSM5121969 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121969 GSM5121969 GEO Accession GSM5121969 SRX10201734 GSM5121970 SRP308746 SRS8348214 GSM5121970 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121970 GSM5121970 GEO Accession GSM5121970 SRX10201735 GSM5121971 SRP308746 SRS8348216 GSM5121971 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121971 GSM5121971 GEO Accession GSM5121971 SRX10201736 GSM5121972 SRP308746 SRS8348215 GSM5121972 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121972 GSM5121972 GEO Accession GSM5121972 SRX10201737 GSM5121973 SRP308746 SRS8348218 GSM5121973 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121973 GSM5121973 GEO Accession GSM5121973 SRX10201738 GSM5121974 SRP308746 SRS8348217 GSM5121974 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121974 GSM5121974 GEO Accession GSM5121974 SRX10201739 GSM5121975 SRP308746 SRS8348219 GSM5121975 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121975 GSM5121975 GEO Accession GSM5121975 SRX10201740 GSM5121976 SRP308746 SRS8348220 GSM5121976 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121976 GSM5121976 GEO Accession GSM5121976 SRX10201741 GSM5121977 SRP308746 SRS8348221 GSM5121977 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121977 GSM5121977 GEO Accession GSM5121977 SRX10201742 GSM5121978 SRP308746 SRS8348222 GSM5121978 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121978 GSM5121978 GEO Accession GSM5121978 SRX10201743 GSM5121979 SRP308746 SRS8348223 GSM5121979 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121979 GSM5121979 GEO Accession GSM5121979 SRX10201744 GSM5121980 SRP308746 SRS8348224 GSM5121980 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121980 GSM5121980 GEO Accession GSM5121980 SRX10201745 GSM5121981 SRP308746 SRS8348225 GSM5121981 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121981 GSM5121981 GEO Accession GSM5121981 SRX10201746 GSM5121982 SRP308746 SRS8348226 GSM5121982 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121982 GSM5121982 GEO Accession GSM5121982 SRX10201747 GSM5121983 SRP308746 SRS8348227 GSM5121983 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121983 GSM5121983 GEO Accession GSM5121983 SRX10201748 GSM5121984 SRP308746 SRS8348228 GSM5121984 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121984 GSM5121984 GEO Accession GSM5121984 SRX10201749 GSM5121985 SRP308746 SRS8348229 GSM5121985 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121985 GSM5121985 GEO Accession GSM5121985 SRX10201750 GSM5121986 SRP308746 SRS8348231 GSM5121986 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121986 GSM5121986 GEO Accession GSM5121986 SRX10201751 GSM5121987 SRP308746 SRS8348232 GSM5121987 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121987 GSM5121987 GEO Accession GSM5121987 SRX10201752 GSM5121988 SRP308746 SRS8348230 GSM5121988 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121988 GSM5121988 GEO Accession GSM5121988 SRX10201753 GSM5121989 SRP308746 SRS8348235 GSM5121989 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121989 GSM5121989 GEO Accession GSM5121989 SRX10201754 GSM5121990 SRP308746 SRS8348233 GSM5121990 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121990 GSM5121990 GEO Accession GSM5121990 SRX10201755 GSM5121991 SRP308746 SRS8348234 GSM5121991 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121991 GSM5121991 GEO Accession GSM5121991 SRX10201756 GSM5121992 SRP308746 SRS8348236 GSM5121992 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121992 GSM5121992 GEO Accession GSM5121992 SRX10201757 GSM5121993 SRP308746 SRS8348237 GSM5121993 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121993 GSM5121993 GEO Accession GSM5121993 SRX10201758 GSM5121994 SRP308746 SRS8348239 GSM5121994 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121994 GSM5121994 GEO Accession GSM5121994 SRX10201759 GSM5121995 SRP308746 SRS8348240 GSM5121995 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121995 GSM5121995 GEO Accession GSM5121995 SRX10201760 GSM5121996 SRP308746 SRS8348238 GSM5121996 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121996 GSM5121996 GEO Accession GSM5121996 SRX10201761 GSM5121997 SRP308746 SRS8348241 GSM5121997 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121997 GSM5121997 GEO Accession GSM5121997 SRX10201762 GSM5121998 SRP308746 SRS8348243 GSM5121998 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121998 GSM5121998 GEO Accession GSM5121998 SRX10201763 GSM5121999 SRP308746 SRS8348242 GSM5121999 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305121999 GSM5121999 GEO Accession GSM5121999 SRX10201764 GSM5122000 SRP308746 SRS8348246 GSM5122000 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122000 GSM5122000 GEO Accession GSM5122000 SRX10201765 GSM5122001 SRP308746 SRS8348244 GSM5122001 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122001 GSM5122001 GEO Accession GSM5122001 SRX10201766 GSM5122002 SRP308746 SRS8348245 GSM5122002 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122002 GSM5122002 GEO Accession GSM5122002 SRX10201767 GSM5122003 SRP308746 SRS8348247 GSM5122003 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122003 GSM5122003 GEO Accession GSM5122003 SRX10201768 GSM5122004 SRP308746 SRS8348250 GSM5122004 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122004 GSM5122004 GEO Accession GSM5122004 SRX10201769 GSM5122005 SRP308746 SRS8348248 GSM5122005 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122005 GSM5122005 GEO Accession GSM5122005 SRX10201770 GSM5122006 SRP308746 SRS8348249 GSM5122006 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122006 GSM5122006 GEO Accession GSM5122006 SRX10201771 GSM5122007 SRP308746 SRS8348251 GSM5122007 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122007 GSM5122007 GEO Accession GSM5122007 SRX10201772 GSM5122008 SRP308746 SRS8348252 GSM5122008 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122008 GSM5122008 GEO Accession GSM5122008 SRX10201773 GSM5122009 SRP308746 SRS8348253 GSM5122009 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122009 GSM5122009 GEO Accession GSM5122009 SRX10201774 GSM5122010 SRP308746 SRS8348256 GSM5122010 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122010 GSM5122010 GEO Accession GSM5122010 SRX10201775 GSM5122011 SRP308746 SRS8348254 GSM5122011 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122011 GSM5122011 GEO Accession GSM5122011 SRX10201776 GSM5122012 SRP308746 SRS8348255 GSM5122012 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122012 GSM5122012 GEO Accession GSM5122012 SRX10201777 GSM5122013 SRP308746 SRS8348257 GSM5122013 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122013 GSM5122013 GEO Accession GSM5122013 SRX10201778 GSM5122014 SRP308746 SRS8348258 GSM5122014 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122014 GSM5122014 GEO Accession GSM5122014 SRX10201779 GSM5122015 SRP308746 SRS8348259 GSM5122015 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122015 GSM5122015 GEO Accession GSM5122015 SRX10201780 GSM5122016 SRP308746 SRS8348260 GSM5122016 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122016 GSM5122016 GEO Accession GSM5122016 SRX10201781 GSM5122017 SRP308746 SRS8348261 GSM5122017 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122017 GSM5122017 GEO Accession GSM5122017 SRX10201782 GSM5122018 SRP308746 SRS8348264 GSM5122018 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122018 GSM5122018 GEO Accession GSM5122018 SRX10201783 GSM5122019 SRP308746 SRS8348262 GSM5122019 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122019 GSM5122019 GEO Accession GSM5122019 SRX10201784 GSM5122020 SRP308746 SRS8348263 GSM5122020 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122020 GSM5122020 GEO Accession GSM5122020 SRX10201785 GSM5122021 SRP308746 SRS8348265 GSM5122021 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122021 GSM5122021 GEO Accession GSM5122021 SRX10201786 GSM5122022 SRP308746 SRS8348268 GSM5122022 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122022 GSM5122022 GEO Accession GSM5122022 SRX10201787 GSM5122023 SRP308746 SRS8348267 GSM5122023 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122023 GSM5122023 GEO Accession GSM5122023 SRX10201788 GSM5122024 SRP308746 SRS8348266 GSM5122024 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122024 GSM5122024 GEO Accession GSM5122024 SRX10201789 GSM5122025 SRP308746 SRS8348269 GSM5122025 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122025 GSM5122025 GEO Accession GSM5122025 SRX10201790 GSM5122026 SRP308746 SRS8348270 GSM5122026 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122026 GSM5122026 GEO Accession GSM5122026 SRX10201791 GSM5122027 SRP308746 SRS8348272 GSM5122027 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122027 GSM5122027 GEO Accession GSM5122027 SRX10201792 GSM5122028 SRP308746 SRS8348271 GSM5122028 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122028 GSM5122028 GEO Accession GSM5122028 SRX10201793 GSM5122029 SRP308746 SRS8348273 GSM5122029 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122029 GSM5122029 GEO Accession GSM5122029 SRX10201794 GSM5122030 SRP308746 SRS8348274 GSM5122030 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122030 GSM5122030 GEO Accession GSM5122030 SRX10201795 GSM5122031 SRP308746 SRS8348275 GSM5122031 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122031 GSM5122031 GEO Accession GSM5122031 SRX10201796 GSM5122032 SRP308746 SRS8348276 GSM5122032 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122032 GSM5122032 GEO Accession GSM5122032 SRX10201797 GSM5122033 SRP308746 SRS8348278 GSM5122033 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122033 GSM5122033 GEO Accession GSM5122033 SRX10201798 GSM5122034 SRP308746 SRS8348277 GSM5122034 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122034 GSM5122034 GEO Accession GSM5122034 SRX10201799 GSM5122035 SRP308746 SRS8348279 GSM5122035 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122035 GSM5122035 GEO Accession GSM5122035 SRX10201800 GSM5122036 SRP308746 SRS8348280 GSM5122036 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122036 GSM5122036 GEO Accession GSM5122036 SRX10201801 GSM5122037 SRP308746 SRS8348282 GSM5122037 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122037 GSM5122037 GEO Accession GSM5122037 SRX10201802 GSM5122038 SRP308746 SRS8348281 GSM5122038 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122038 GSM5122038 GEO Accession GSM5122038 SRX10201803 GSM5122039 SRP308746 SRS8348284 GSM5122039 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122039 GSM5122039 GEO Accession GSM5122039 SRX10201804 GSM5122040 SRP308746 SRS8348283 GSM5122040 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122040 GSM5122040 GEO Accession GSM5122040 SRX10201805 GSM5122041 SRP308746 SRS8348285 GSM5122041 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122041 GSM5122041 GEO Accession GSM5122041 SRX10201806 GSM5122042 SRP308746 SRS8348286 GSM5122042 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122042 GSM5122042 GEO Accession GSM5122042 SRX10201807 GSM5122043 SRP308746 SRS8348287 GSM5122043 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122043 GSM5122043 GEO Accession GSM5122043 SRX10201808 GSM5122044 SRP308746 SRS8348288 GSM5122044 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122044 GSM5122044 GEO Accession GSM5122044 SRX10201809 GSM5122045 SRP308746 SRS8348289 GSM5122045 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122045 GSM5122045 GEO Accession GSM5122045 SRX10201810 GSM5122046 SRP308746 SRS8348290 GSM5122046 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122046 GSM5122046 GEO Accession GSM5122046 SRX10201811 GSM5122049 SRP308746 SRS8348291 GSM5122049 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122049 GSM5122049 GEO Accession GSM5122049 SRX10201812 GSM5122051 SRP308746 SRS8348292 GSM5122051 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122051 GSM5122051 GEO Accession GSM5122051 SRX10201813 GSM5122055 SRP308746 SRS8348293 GSM5122055 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122055 GSM5122055 GEO Accession GSM5122055 SRX10201815 GSM5122061 SRP308746 SRS8348296 GSM5122061 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122061 GSM5122061 GEO Accession GSM5122061 SRX10201816 GSM5122065 SRP308746 SRS8348295 GSM5122065 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122065 GSM5122065 GEO Accession GSM5122065 SRX10201817 GSM5122068 SRP308746 SRS8348298 GSM5122068 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122068 GSM5122068 GEO Accession GSM5122068 SRX10201818 GSM5122072 SRP308746 SRS8348297 GSM5122072 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122072 GSM5122072 GEO Accession GSM5122072 SRX10201819 GSM5122076 SRP308746 SRS8348300 GSM5122076 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122076 GSM5122076 GEO Accession GSM5122076 SRX10201820 GSM5122079 SRP308746 SRS8348299 GSM5122079 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122079 GSM5122079 GEO Accession GSM5122079 SRX10201821 GSM5122083 SRP308746 SRS8348302 GSM5122083 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122083 GSM5122083 GEO Accession GSM5122083 SRX10201822 GSM5122086 SRP308746 SRS8348301 GSM5122086 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122086 GSM5122086 GEO Accession GSM5122086 SRX10201823 GSM5122087 SRP308746 SRS8348303 GSM5122087 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122087 GSM5122087 GEO Accession GSM5122087 SRX10201824 GSM5122088 SRP308746 SRS8348304 GSM5122088 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122088 GSM5122088 GEO Accession GSM5122088 SRX10201825 GSM5122089 SRP308746 SRS8348305 GSM5122089 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122089 GSM5122089 GEO Accession GSM5122089 SRX10201826 GSM5122090 SRP308746 SRS8348306 GSM5122090 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122090 GSM5122090 GEO Accession GSM5122090 SRX10201827 GSM5122091 SRP308746 SRS8348307 GSM5122091 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122091 GSM5122091 GEO Accession GSM5122091 SRX10201828 GSM5122092 SRP308746 SRS8348309 GSM5122092 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122092 GSM5122092 GEO Accession GSM5122092 SRX10201829 GSM5122093 SRP308746 SRS8348308 GSM5122093 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122093 GSM5122093 GEO Accession GSM5122093 SRX10201830 GSM5122094 SRP308746 SRS8348310 GSM5122094 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122094 GSM5122094 GEO Accession GSM5122094 SRX10201831 GSM5122095 SRP308746 SRS8348311 GSM5122095 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122095 GSM5122095 GEO Accession GSM5122095 SRX10201832 GSM5122096 SRP308746 SRS8348312 GSM5122096 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122096 GSM5122096 GEO Accession GSM5122096 SRX10201833 GSM5122097 SRP308746 SRS8348313 GSM5122097 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122097 GSM5122097 GEO Accession GSM5122097 SRX10201834 GSM5122098 SRP308746 SRS8348314 GSM5122098 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122098 GSM5122098 GEO Accession GSM5122098 SRX10201835 GSM5122099 SRP308746 SRS8348315 GSM5122099 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122099 GSM5122099 GEO Accession GSM5122099 SRX10201836 GSM5122100 SRP308746 SRS8348316 GSM5122100 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122100 GSM5122100 GEO Accession GSM5122100 SRX10201837 GSM5122101 SRP308746 SRS8348317 GSM5122101 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122101 GSM5122101 GEO Accession GSM5122101 SRX10201838 GSM5122102 SRP308746 SRS8348318 GSM5122102 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122102 GSM5122102 GEO Accession GSM5122102 SRX10201839 GSM5122103 SRP308746 SRS8348319 GSM5122103 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122103 GSM5122103 GEO Accession GSM5122103 SRX10201840 GSM5122104 SRP308746 SRS8348320 GSM5122104 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122104 GSM5122104 GEO Accession GSM5122104 SRX10201841 GSM5122105 SRP308746 SRS8348321 GSM5122105 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122105 GSM5122105 GEO Accession GSM5122105 SRX10201842 GSM5122106 SRP308746 SRS8348322 GSM5122106 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122106 GSM5122106 GEO Accession GSM5122106 SRX10201843 GSM5122107 SRP308746 SRS8348323 GSM5122107 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122107 GSM5122107 GEO Accession GSM5122107 SRX10201844 GSM5122108 SRP308746 SRS8348324 GSM5122108 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122108 GSM5122108 GEO Accession GSM5122108 SRX10201845 GSM5122109 SRP308746 SRS8348325 GSM5122109 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122109 GSM5122109 GEO Accession GSM5122109 SRX10201846 GSM5122110 SRP308746 SRS8348326 GSM5122110 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122110 GSM5122110 GEO Accession GSM5122110 SRX10201847 GSM5122111 SRP308746 SRS8348328 GSM5122111 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122111 GSM5122111 GEO Accession GSM5122111 SRX10201848 GSM5122112 SRP308746 SRS8348327 GSM5122112 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122112 GSM5122112 GEO Accession GSM5122112 SRX10201849 GSM5122113 SRP308746 SRS8348329 GSM5122113 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122113 GSM5122113 GEO Accession GSM5122113 SRX10201850 GSM5122114 SRP308746 SRS8348330 GSM5122114 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122114 GSM5122114 GEO Accession GSM5122114 SRX10201851 GSM5122115 SRP308746 SRS8348331 GSM5122115 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122115 GSM5122115 GEO Accession GSM5122115 SRX10201852 GSM5122116 SRP308746 SRS8348332 GSM5122116 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122116 GSM5122116 GEO Accession GSM5122116 SRX10201853 GSM5122117 SRP308746 SRS8348333 GSM5122117 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122117 GSM5122117 GEO Accession GSM5122117 SRX10201854 GSM5122118 SRP308746 SRS8348334 GSM5122118 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122118 GSM5122118 GEO Accession GSM5122118 SRX10201855 GSM5122119 SRP308746 SRS8348335 GSM5122119 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122119 GSM5122119 GEO Accession GSM5122119 SRX10201856 GSM5122120 SRP308746 SRS8348337 GSM5122120 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122120 GSM5122120 GEO Accession GSM5122120 SRX10201857 GSM5122121 SRP308746 SRS8348336 GSM5122121 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122121 GSM5122121 GEO Accession GSM5122121 SRX10201858 GSM5122122 SRP308746 SRS8348338 GSM5122122 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122122 GSM5122122 GEO Accession GSM5122122 SRX10201859 GSM5122123 SRP308746 SRS8348340 GSM5122123 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122123 GSM5122123 GEO Accession GSM5122123 SRX10201860 GSM5122124 SRP308746 SRS8348339 GSM5122124 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122124 GSM5122124 GEO Accession GSM5122124 SRX10201861 GSM5122125 SRP308746 SRS8348344 GSM5122125 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122125 GSM5122125 GEO Accession GSM5122125 SRX10201862 GSM5122126 SRP308746 SRS8348342 GSM5122126 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122126 GSM5122126 GEO Accession GSM5122126 SRX10201863 GSM5122127 SRP308746 SRS8348341 GSM5122127 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122127 GSM5122127 GEO Accession GSM5122127 SRX10201864 GSM5122128 SRP308746 SRS8348343 GSM5122128 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122128 GSM5122128 GEO Accession GSM5122128 SRX10201865 GSM5122129 SRP308746 SRS8348345 GSM5122129 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122129 GSM5122129 GEO Accession GSM5122129 SRX10201866 GSM5122130 SRP308746 SRS8348346 GSM5122130 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122130 GSM5122130 GEO Accession GSM5122130 SRX10201867 GSM5122131 SRP308746 SRS8348347 GSM5122131 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122131 GSM5122131 GEO Accession GSM5122131 SRX10201868 GSM5122132 SRP308746 SRS8348348 GSM5122132 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122132 GSM5122132 GEO Accession GSM5122132 SRX10201869 GSM5122133 SRP308746 SRS8348349 GSM5122133 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122133 GSM5122133 GEO Accession GSM5122133 SRX10201870 GSM5122134 SRP308746 SRS8348350 GSM5122134 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122134 GSM5122134 GEO Accession GSM5122134 SRX10201871 GSM5122135 SRP308746 SRS8348352 GSM5122135 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122135 GSM5122135 GEO Accession GSM5122135 SRX10201872 GSM5122136 SRP308746 SRS8348351 GSM5122136 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122136 GSM5122136 GEO Accession GSM5122136 SRX10201873 GSM5122137 SRP308746 SRS8348353 GSM5122137 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122137 GSM5122137 GEO Accession GSM5122137 SRX10201874 GSM5122138 SRP308746 SRS8348354 GSM5122138 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122138 GSM5122138 GEO Accession GSM5122138 SRX10201875 GSM5122139 SRP308746 SRS8348356 GSM5122139 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122139 GSM5122139 GEO Accession GSM5122139 SRX10201876 GSM5122140 SRP308746 SRS8348357 GSM5122140 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122140 GSM5122140 GEO Accession GSM5122140 SRX10201877 GSM5122141 SRP308746 SRS8348355 GSM5122141 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122141 GSM5122141 GEO Accession GSM5122141 SRX10201878 GSM5122142 SRP308746 SRS8348358 GSM5122142 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122142 GSM5122142 GEO Accession GSM5122142 SRX10201879 GSM5122143 SRP308746 SRS8348360 GSM5122143 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122143 GSM5122143 GEO Accession GSM5122143 SRX10201880 GSM5122144 SRP308746 SRS8348359 GSM5122144 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122144 GSM5122144 GEO Accession GSM5122144 SRX10201881 GSM5122145 SRP308746 SRS8348361 GSM5122145 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122145 GSM5122145 GEO Accession GSM5122145 SRX10201882 GSM5122146 SRP308746 SRS8348362 GSM5122146 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122146 GSM5122146 GEO Accession GSM5122146 SRX10201883 GSM5122147 SRP308746 SRS8348363 GSM5122147 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122147 GSM5122147 GEO Accession GSM5122147 SRX10201884 GSM5122148 SRP308746 SRS8348364 GSM5122148 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122148 GSM5122148 GEO Accession GSM5122148 SRX10201885 GSM5122149 SRP308746 SRS8348365 GSM5122149 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122149 GSM5122149 GEO Accession GSM5122149 SRX10201886 GSM5122150 SRP308746 SRS8348366 GSM5122150 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122150 GSM5122150 GEO Accession GSM5122150 SRX10201887 GSM5122151 SRP308746 SRS8348367 GSM5122151 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122151 GSM5122151 GEO Accession GSM5122151 SRX10201888 GSM5122152 SRP308746 SRS8348369 GSM5122152 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122152 GSM5122152 GEO Accession GSM5122152 SRX10201889 GSM5122153 SRP308746 SRS8348368 GSM5122153 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122153 GSM5122153 GEO Accession GSM5122153 SRX10201890 GSM5122154 SRP308746 SRS8348370 GSM5122154 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122154 GSM5122154 GEO Accession GSM5122154 SRX10201891 GSM5122155 SRP308746 SRS8348372 GSM5122155 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122155 GSM5122155 GEO Accession GSM5122155 SRX10201892 GSM5122156 SRP308746 SRS8348371 GSM5122156 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122156 GSM5122156 GEO Accession GSM5122156 SRX10201893 GSM5122157 SRP308746 SRS8348373 GSM5122157 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122157 GSM5122157 GEO Accession GSM5122157 SRX10201894 GSM5122158 SRP308746 SRS8348374 GSM5122158 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122158 GSM5122158 GEO Accession GSM5122158 SRX10201895 GSM5122159 SRP308746 SRS8348375 GSM5122159 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122159 GSM5122159 GEO Accession GSM5122159 SRX10201896 GSM5122160 SRP308746 SRS8348376 GSM5122160 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122160 GSM5122160 GEO Accession GSM5122160 SRX10201897 GSM5122161 SRP308746 SRS8348377 GSM5122161 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122161 GSM5122161 GEO Accession GSM5122161 SRX10201898 GSM5122162 SRP308746 SRS8348378 GSM5122162 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122162 GSM5122162 GEO Accession GSM5122162 SRX10201899 GSM5122163 SRP308746 SRS8348379 GSM5122163 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122163 GSM5122163 GEO Accession GSM5122163 SRX10201900 GSM5122164 SRP308746 SRS8348381 GSM5122164 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122164 GSM5122164 GEO Accession GSM5122164 SRX10201901 GSM5122165 SRP308746 SRS8348380 GSM5122165 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122165 GSM5122165 GEO Accession GSM5122165 SRX10201902 GSM5122166 SRP308746 SRS8348382 GSM5122166 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122166 GSM5122166 GEO Accession GSM5122166 SRX10201903 GSM5122167 SRP308746 SRS8348383 GSM5122167 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122167 GSM5122167 GEO Accession GSM5122167 SRX10201904 GSM5122168 SRP308746 SRS8348384 GSM5122168 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122168 GSM5122168 GEO Accession GSM5122168 SRX10201905 GSM5122169 SRP308746 SRS8348385 GSM5122169 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122169 GSM5122169 GEO Accession GSM5122169 SRX10201906 GSM5122170 SRP308746 SRS8348386 GSM5122170 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122170 GSM5122170 GEO Accession GSM5122170 SRX10201907 GSM5122171 SRP308746 SRS8348387 GSM5122171 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122171 GSM5122171 GEO Accession GSM5122171 SRX10201908 GSM5122172 SRP308746 SRS8348389 GSM5122172 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122172 GSM5122172 GEO Accession GSM5122172 SRX10201909 GSM5122173 SRP308746 SRS8348388 GSM5122173 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122173 GSM5122173 GEO Accession GSM5122173 SRX10201910 GSM5122174 SRP308746 SRS8348391 GSM5122174 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122174 GSM5122174 GEO Accession GSM5122174 SRX10201911 GSM5122175 SRP308746 SRS8348390 GSM5122175 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122175 GSM5122175 GEO Accession GSM5122175 SRX10201912 GSM5122176 SRP308746 SRS8348392 GSM5122176 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122176 GSM5122176 GEO Accession GSM5122176 SRX10201913 GSM5122177 SRP308746 SRS8348393 GSM5122177 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122177 GSM5122177 GEO Accession GSM5122177 SRX10201914 GSM5122178 SRP308746 SRS8348395 GSM5122178 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122178 GSM5122178 GEO Accession GSM5122178 SRX10201915 GSM5122179 SRP308746 SRS8348394 GSM5122179 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122179 GSM5122179 GEO Accession GSM5122179 SRX10201916 GSM5122180 SRP308746 SRS8348396 GSM5122180 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122180 GSM5122180 GEO Accession GSM5122180 SRX10201917 GSM5122181 SRP308746 SRS8348397 GSM5122181 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122181 GSM5122181 GEO Accession GSM5122181 SRX10201918 GSM5122182 SRP308746 SRS8348399 GSM5122182 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122182 GSM5122182 GEO Accession GSM5122182 SRX10201919 GSM5122183 SRP308746 SRS8348398 GSM5122183 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122183 GSM5122183 GEO Accession GSM5122183 SRX10201920 GSM5122184 SRP308746 SRS8348400 GSM5122184 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122184 GSM5122184 GEO Accession GSM5122184 SRX10201921 GSM5122185 SRP308746 SRS8348401 GSM5122185 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122185 GSM5122185 GEO Accession GSM5122185 SRX10201922 GSM5122186 SRP308746 SRS8348402 GSM5122186 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122186 GSM5122186 GEO Accession GSM5122186 SRX10201923 GSM5122187 SRP308746 SRS8348404 GSM5122187 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122187 GSM5122187 GEO Accession GSM5122187 SRX10201924 GSM5122188 SRP308746 SRS8348403 GSM5122188 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122188 GSM5122188 GEO Accession GSM5122188 SRX10201925 GSM5122189 SRP308746 SRS8348405 GSM5122189 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122189 GSM5122189 GEO Accession GSM5122189 SRX10201926 GSM5122190 SRP308746 SRS8348407 GSM5122190 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122190 GSM5122190 GEO Accession GSM5122190 SRX10201927 GSM5122191 SRP308746 SRS8348406 GSM5122191 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122191 GSM5122191 GEO Accession GSM5122191 SRX10201928 GSM5122192 SRP308746 SRS8348410 GSM5122192 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122192 GSM5122192 GEO Accession GSM5122192 SRX10201929 GSM5122193 SRP308746 SRS8348408 GSM5122193 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122193 GSM5122193 GEO Accession GSM5122193 SRX10201930 GSM5122194 SRP308746 SRS8348409 GSM5122194 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122194 GSM5122194 GEO Accession GSM5122194 SRX10201931 GSM5122195 SRP308746 SRS8348412 GSM5122195 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122195 GSM5122195 GEO Accession GSM5122195 SRX10201932 GSM5122196 SRP308746 SRS8348411 GSM5122196 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122196 GSM5122196 GEO Accession GSM5122196 SRX10201933 GSM5122197 SRP308746 SRS8348414 GSM5122197 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122197 GSM5122197 GEO Accession GSM5122197 SRX10201934 GSM5122198 SRP308746 SRS8348413 GSM5122198 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122198 GSM5122198 GEO Accession GSM5122198 SRX10201935 GSM5122199 SRP308746 SRS8348416 GSM5122199 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122199 GSM5122199 GEO Accession GSM5122199 SRX10201936 GSM5122200 SRP308746 SRS8348415 GSM5122200 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122200 GSM5122200 GEO Accession GSM5122200 SRX10201937 GSM5122201 SRP308746 SRS8348417 GSM5122201 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122201 GSM5122201 GEO Accession GSM5122201 SRX10201938 GSM5122202 SRP308746 SRS8348418 GSM5122202 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122202 GSM5122202 GEO Accession GSM5122202 SRX10201939 GSM5122203 SRP308746 SRS8348419 GSM5122203 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122203 GSM5122203 GEO Accession GSM5122203 SRX10201940 GSM5122204 SRP308746 SRS8348421 GSM5122204 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122204 GSM5122204 GEO Accession GSM5122204 SRX10201941 GSM5122205 SRP308746 SRS8348420 GSM5122205 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122205 GSM5122205 GEO Accession GSM5122205 SRX10201942 GSM5122206 SRP308746 SRS8348423 GSM5122206 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122206 GSM5122206 GEO Accession GSM5122206 SRX10201943 GSM5122207 SRP308746 SRS8348422 GSM5122207 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122207 GSM5122207 GEO Accession GSM5122207 SRX10201944 GSM5122208 SRP308746 SRS8348424 GSM5122208 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122208 GSM5122208 GEO Accession GSM5122208 SRX10201945 GSM5122209 SRP308746 SRS8348425 GSM5122209 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122209 GSM5122209 GEO Accession GSM5122209 SRX10201946 GSM5122210 SRP308746 SRS8348426 GSM5122210 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122210 GSM5122210 GEO Accession GSM5122210 SRX10201947 GSM5122211 SRP308746 SRS8348429 GSM5122211 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122211 GSM5122211 GEO Accession GSM5122211 SRX10201948 GSM5122212 SRP308746 SRS8348430 GSM5122212 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122212 GSM5122212 GEO Accession GSM5122212 SRX10201949 GSM5122213 SRP308746 SRS8348427 GSM5122213 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122213 GSM5122213 GEO Accession GSM5122213 SRX10201950 GSM5122214 SRP308746 SRS8348428 GSM5122214 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122214 GSM5122214 GEO Accession GSM5122214 SRX10201951 GSM5122215 SRP308746 SRS8348431 GSM5122215 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122215 GSM5122215 GEO Accession GSM5122215 SRX10201952 GSM5122216 SRP308746 SRS8348432 GSM5122216 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122216 GSM5122216 GEO Accession GSM5122216 SRX10201953 GSM5122217 SRP308746 SRS8348433 GSM5122217 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122217 GSM5122217 GEO Accession GSM5122217 SRX10201954 GSM5122218 SRP308746 SRS8348436 GSM5122218 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122218 GSM5122218 GEO Accession GSM5122218 SRX10201955 GSM5122219 SRP308746 SRS8348435 GSM5122219 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122219 GSM5122219 GEO Accession GSM5122219 SRX10201956 GSM5122220 SRP308746 SRS8348434 GSM5122220 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122220 GSM5122220 GEO Accession GSM5122220 SRX10201957 GSM5122221 SRP308746 SRS8348440 GSM5122221 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122221 GSM5122221 GEO Accession GSM5122221 SRX10201958 GSM5122222 SRP308746 SRS8348437 GSM5122222 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122222 GSM5122222 GEO Accession GSM5122222 SRX10201959 GSM5122223 SRP308746 SRS8348438 GSM5122223 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122223 GSM5122223 GEO Accession GSM5122223 SRX10201960 GSM5122224 SRP308746 SRS8348439 GSM5122224 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122224 GSM5122224 GEO Accession GSM5122224 SRX10201961 GSM5122225 SRP308746 SRS8348442 GSM5122225 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122225 GSM5122225 GEO Accession GSM5122225 SRX10201962 GSM5122226 SRP308746 SRS8348443 GSM5122226 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122226 GSM5122226 GEO Accession GSM5122226 SRX10201963 GSM5122227 SRP308746 SRS8348441 GSM5122227 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122227 GSM5122227 GEO Accession GSM5122227 SRX10201964 GSM5122228 SRP308746 SRS8348444 GSM5122228 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122228 GSM5122228 GEO Accession GSM5122228 SRX10201965 GSM5122229 SRP308746 SRS8348445 GSM5122229 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122229 GSM5122229 GEO Accession GSM5122229 SRX10201966 GSM5122230 SRP308746 SRS8348447 GSM5122230 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122230 GSM5122230 GEO Accession GSM5122230 SRX10201967 GSM5122231 SRP308746 SRS8348446 GSM5122231 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122231 GSM5122231 GEO Accession GSM5122231 SRX10201968 GSM5122232 SRP308746 SRS8348448 GSM5122232 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122232 GSM5122232 GEO Accession GSM5122232 SRX10201969 GSM5122233 SRP308746 SRS8348449 GSM5122233 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122233 GSM5122233 GEO Accession GSM5122233 SRX10201970 GSM5122234 SRP308746 SRS8348450 GSM5122234 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122234 GSM5122234 GEO Accession GSM5122234 SRX10201971 GSM5122235 SRP308746 SRS8348452 GSM5122235 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122235 GSM5122235 GEO Accession GSM5122235 SRX10201972 GSM5122236 SRP308746 SRS8348451 GSM5122236 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122236 GSM5122236 GEO Accession GSM5122236 SRX10201973 GSM5122237 SRP308746 SRS8348454 GSM5122237 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122237 GSM5122237 GEO Accession GSM5122237 SRX10201974 GSM5122238 SRP308746 SRS8348453 GSM5122238 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122238 GSM5122238 GEO Accession GSM5122238 SRX10201975 GSM5122239 SRP308746 SRS8348455 GSM5122239 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122239 GSM5122239 GEO Accession GSM5122239 SRX10201976 GSM5122240 SRP308746 SRS8348456 GSM5122240 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122240 GSM5122240 GEO Accession GSM5122240 SRX10201977 GSM5122241 SRP308746 SRS8348458 GSM5122241 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122241 GSM5122241 GEO Accession GSM5122241 SRX10201978 GSM5122242 SRP308746 SRS8348457 GSM5122242 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122242 GSM5122242 GEO Accession GSM5122242 SRX10201979 GSM5122243 SRP308746 SRS8348459 GSM5122243 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122243 GSM5122243 GEO Accession GSM5122243 SRX10201980 GSM5122244 SRP308746 SRS8348460 GSM5122244 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122244 GSM5122244 GEO Accession GSM5122244 SRX10201981 GSM5122245 SRP308746 SRS8348462 GSM5122245 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122245 GSM5122245 GEO Accession GSM5122245 SRX10201982 GSM5122246 SRP308746 SRS8348461 GSM5122246 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122246 GSM5122246 GEO Accession GSM5122246 SRX10201983 GSM5122247 SRP308746 SRS8348463 GSM5122247 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122247 GSM5122247 GEO Accession GSM5122247 SRX10201984 GSM5122248 SRP308746 SRS8348464 GSM5122248 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122248 GSM5122248 GEO Accession GSM5122248 SRX10201985 GSM5122249 SRP308746 SRS8348466 GSM5122249 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122249 GSM5122249 GEO Accession GSM5122249 SRX10201986 GSM5122250 SRP308746 SRS8348465 GSM5122250 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122250 GSM5122250 GEO Accession GSM5122250 SRX10201987 GSM5122251 SRP308746 SRS8348467 GSM5122251 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122251 GSM5122251 GEO Accession GSM5122251 SRX10201988 GSM5122252 SRP308746 SRS8348468 GSM5122252 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122252 GSM5122252 GEO Accession GSM5122252 SRX10201989 GSM5122253 SRP308746 SRS8348469 GSM5122253 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122253 GSM5122253 GEO Accession GSM5122253 SRX10201990 GSM5122254 SRP308746 SRS8348470 GSM5122254 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122254 GSM5122254 GEO Accession GSM5122254 SRX10201991 GSM5122255 SRP308746 SRS8348471 GSM5122255 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122255 GSM5122255 GEO Accession GSM5122255 SRX10201992 GSM5122256 SRP308746 SRS8348472 GSM5122256 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122256 GSM5122256 GEO Accession GSM5122256 SRX10201993 GSM5122257 SRP308746 SRS8348474 GSM5122257 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122257 GSM5122257 GEO Accession GSM5122257 SRX10201994 GSM5122258 SRP308746 SRS8348475 GSM5122258 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122258 GSM5122258 GEO Accession GSM5122258 SRX10201995 GSM5122259 SRP308746 SRS8348473 GSM5122259 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122259 GSM5122259 GEO Accession GSM5122259 SRX10201996 GSM5122260 SRP308746 SRS8348476 GSM5122260 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122260 GSM5122260 GEO Accession GSM5122260 SRX10201997 GSM5122261 SRP308746 SRS8348478 GSM5122261 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122261 GSM5122261 GEO Accession GSM5122261 SRX10201998 GSM5122262 SRP308746 SRS8348477 GSM5122262 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122262 GSM5122262 GEO Accession GSM5122262 SRX10201999 GSM5122263 SRP308746 SRS8348479 GSM5122263 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122263 GSM5122263 GEO Accession GSM5122263 SRX10202000 GSM5122264 SRP308746 SRS8348480 GSM5122264 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122264 GSM5122264 GEO Accession GSM5122264 SRX10202001 GSM5122265 SRP308746 SRS8348482 GSM5122265 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122265 GSM5122265 GEO Accession GSM5122265 SRX10202002 GSM5122266 SRP308746 SRS8348481 GSM5122266 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122266 GSM5122266 GEO Accession GSM5122266 SRX10202003 GSM5122267 SRP308746 SRS8348483 GSM5122267 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122267 GSM5122267 GEO Accession GSM5122267 SRX10202004 GSM5122268 SRP308746 SRS8348484 GSM5122268 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122268 GSM5122268 GEO Accession GSM5122268 SRX10202005 GSM5122269 SRP308746 SRS8348485 GSM5122269 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122269 GSM5122269 GEO Accession GSM5122269 SRX10202006 GSM5122270 SRP308746 SRS8348486 GSM5122270 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122270 GSM5122270 GEO Accession GSM5122270 SRX10202007 GSM5122271 SRP308746 SRS8348488 GSM5122271 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122271 GSM5122271 GEO Accession GSM5122271 SRX10202008 GSM5122272 SRP308746 SRS8348487 GSM5122272 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122272 GSM5122272 GEO Accession GSM5122272 SRX10202009 GSM5122273 SRP308746 SRS8348489 GSM5122273 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122273 GSM5122273 GEO Accession GSM5122273 SRX10202010 GSM5122274 SRP308746 SRS8348490 GSM5122274 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122274 GSM5122274 GEO Accession GSM5122274 SRX10202011 GSM5122275 SRP308746 SRS8348491 GSM5122275 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122275 GSM5122275 GEO Accession GSM5122275 SRX10202012 GSM5122276 SRP308746 SRS8348492 GSM5122276 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122276 GSM5122276 GEO Accession GSM5122276 SRX10202013 GSM5122277 SRP308746 SRS8348493 GSM5122277 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122277 GSM5122277 GEO Accession GSM5122277 SRX10202014 GSM5122278 SRP308746 SRS8348494 GSM5122278 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122278 GSM5122278 GEO Accession GSM5122278 SRX10202015 GSM5122279 SRP308746 SRS8348495 GSM5122279 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122279 GSM5122279 GEO Accession GSM5122279 SRX10202016 GSM5122280 SRP308746 SRS8348497 GSM5122280 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122280 GSM5122280 GEO Accession GSM5122280 SRX10202017 GSM5122281 SRP308746 SRS8348496 GSM5122281 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122281 GSM5122281 GEO Accession GSM5122281 SRX10202018 GSM5122282 SRP308746 SRS8348498 GSM5122282 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122282 GSM5122282 GEO Accession GSM5122282 SRX10202019 GSM5122283 SRP308746 SRS8348499 GSM5122283 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122283 GSM5122283 GEO Accession GSM5122283 SRX10202020 GSM5122284 SRP308746 SRS8348500 GSM5122284 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122284 GSM5122284 GEO Accession GSM5122284 SRX10202021 GSM5122285 SRP308746 SRS8348502 GSM5122285 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122285 GSM5122285 GEO Accession GSM5122285 SRX10202022 GSM5122286 SRP308746 SRS8348501 GSM5122286 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122286 GSM5122286 GEO Accession GSM5122286 SRX10202023 GSM5122287 SRP308746 SRS8348503 GSM5122287 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122287 GSM5122287 GEO Accession GSM5122287 SRX10202024 GSM5122288 SRP308746 SRS8348504 GSM5122288 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122288 GSM5122288 GEO Accession GSM5122288 SRX10202025 GSM5122289 SRP308746 SRS8348505 GSM5122289 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122289 GSM5122289 GEO Accession GSM5122289 SRX10202026 GSM5122290 SRP308746 SRS8348506 GSM5122290 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122290 GSM5122290 GEO Accession GSM5122290 SRX10202027 GSM5122291 SRP308746 SRS8348507 GSM5122291 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122291 GSM5122291 GEO Accession GSM5122291 SRX10202028 GSM5122292 SRP308746 SRS8348509 GSM5122292 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122292 GSM5122292 GEO Accession GSM5122292 SRX10202029 GSM5122293 SRP308746 SRS8348508 GSM5122293 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122293 GSM5122293 GEO Accession GSM5122293 SRX10202030 GSM5122294 SRP308746 SRS8348510 GSM5122294 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122294 GSM5122294 GEO Accession GSM5122294 SRX10202031 GSM5122295 SRP308746 SRS8348511 GSM5122295 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122295 GSM5122295 GEO Accession GSM5122295 SRX10202032 GSM5122296 SRP308746 SRS8348514 GSM5122296 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122296 GSM5122296 GEO Accession GSM5122296 SRX10202033 GSM5122297 SRP308746 SRS8348513 GSM5122297 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122297 GSM5122297 GEO Accession GSM5122297 SRX10202034 GSM5122298 SRP308746 SRS8348512 GSM5122298 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122298 GSM5122298 GEO Accession GSM5122298 SRX10202035 GSM5122299 SRP308746 SRS8348515 GSM5122299 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122299 GSM5122299 GEO Accession GSM5122299 SRX10202036 GSM5122300 SRP308746 SRS8348516 GSM5122300 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122300 GSM5122300 GEO Accession GSM5122300 SRX10202037 GSM5122301 SRP308746 SRS8348517 GSM5122301 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122301 GSM5122301 GEO Accession GSM5122301 SRX10202038 GSM5122302 SRP308746 SRS8348519 GSM5122302 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122302 GSM5122302 GEO Accession GSM5122302 SRX10202039 GSM5122303 SRP308746 SRS8348518 GSM5122303 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122303 GSM5122303 GEO Accession GSM5122303 SRX10202040 GSM5122304 SRP308746 SRS8348520 GSM5122304 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122304 GSM5122304 GEO Accession GSM5122304 SRX10202041 GSM5122305 SRP308746 SRS8348521 GSM5122305 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122305 GSM5122305 GEO Accession GSM5122305 SRX10202042 GSM5122306 SRP308746 SRS8348522 GSM5122306 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122306 GSM5122306 GEO Accession GSM5122306 SRX10202043 GSM5122307 SRP308746 SRS8348523 GSM5122307 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122307 GSM5122307 GEO Accession GSM5122307 SRX10202044 GSM5122308 SRP308746 SRS8348524 GSM5122308 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122308 GSM5122308 GEO Accession GSM5122308 SRX10202045 GSM5122309 SRP308746 SRS8348525 GSM5122309 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122309 GSM5122309 GEO Accession GSM5122309 SRX10202046 GSM5122310 SRP308746 SRS8348526 GSM5122310 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122310 GSM5122310 GEO Accession GSM5122310 SRX10202047 GSM5122311 SRP308746 SRS8348527 GSM5122311 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122311 GSM5122311 GEO Accession GSM5122311 SRX10202048 GSM5122312 SRP308746 SRS8348528 GSM5122312 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122312 GSM5122312 GEO Accession GSM5122312 SRX10202049 GSM5122313 SRP308746 SRS8348529 GSM5122313 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122313 GSM5122313 GEO Accession GSM5122313 SRX10202050 GSM5122314 SRP308746 SRS8348530 GSM5122314 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122314 GSM5122314 GEO Accession GSM5122314 SRX10202051 GSM5122315 SRP308746 SRS8348531 GSM5122315 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122315 GSM5122315 GEO Accession GSM5122315 SRX10202052 GSM5122316 SRP308746 SRS8348532 GSM5122316 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122316 GSM5122316 GEO Accession GSM5122316 SRX10202053 GSM5122317 SRP308746 SRS8348534 GSM5122317 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122317 GSM5122317 GEO Accession GSM5122317 SRX10202054 GSM5122318 SRP308746 SRS8348533 GSM5122318 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122318 GSM5122318 GEO Accession GSM5122318 SRX10202055 GSM5122319 SRP308746 SRS8348535 GSM5122319 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122319 GSM5122319 GEO Accession GSM5122319 SRX10202056 GSM5122320 SRP308746 SRS8348537 GSM5122320 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122320 GSM5122320 GEO Accession GSM5122320 SRX10202057 GSM5122321 SRP308746 SRS8348536 GSM5122321 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122321 GSM5122321 GEO Accession GSM5122321 SRX10202058 GSM5122322 SRP308746 SRS8348538 GSM5122322 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122322 GSM5122322 GEO Accession GSM5122322 SRX10202059 GSM5122323 SRP308746 SRS8348539 GSM5122323 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122323 GSM5122323 GEO Accession GSM5122323 SRX10202060 GSM5122324 SRP308746 SRS8348541 GSM5122324 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122324 GSM5122324 GEO Accession GSM5122324 SRX10202061 GSM5122325 SRP308746 SRS8348540 GSM5122325 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122325 GSM5122325 GEO Accession GSM5122325 SRX10202062 GSM5122326 SRP308746 SRS8348542 GSM5122326 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122326 GSM5122326 GEO Accession GSM5122326 SRX10202063 GSM5122327 SRP308746 SRS8348543 GSM5122327 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122327 GSM5122327 GEO Accession GSM5122327 SRX10202064 GSM5122328 SRP308746 SRS8348544 GSM5122328 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122328 GSM5122328 GEO Accession GSM5122328 SRX10202065 GSM5122329 SRP308746 SRS8348546 GSM5122329 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122329 GSM5122329 GEO Accession GSM5122329 SRX10202066 GSM5122330 SRP308746 SRS8348545 GSM5122330 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122330 GSM5122330 GEO Accession GSM5122330 SRX10202067 GSM5122331 SRP308746 SRS8348547 GSM5122331 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122331 GSM5122331 GEO Accession GSM5122331 SRX10202068 GSM5122332 SRP308746 SRS8348548 GSM5122332 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122332 GSM5122332 GEO Accession GSM5122332 SRX10202069 GSM5122333 SRP308746 SRS8348550 GSM5122333 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122333 GSM5122333 GEO Accession GSM5122333 SRX10202070 GSM5122334 SRP308746 SRS8348549 GSM5122334 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122334 GSM5122334 GEO Accession GSM5122334 SRX10202071 GSM5122335 SRP308746 SRS8348551 GSM5122335 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122335 GSM5122335 GEO Accession GSM5122335 SRX10202072 GSM5122336 SRP308746 SRS8348552 GSM5122336 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122336 GSM5122336 GEO Accession GSM5122336 SRX10202073 GSM5122337 SRP308746 SRS8348554 GSM5122337 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122337 GSM5122337 GEO Accession GSM5122337 SRX10202074 GSM5122338 SRP308746 SRS8348553 GSM5122338 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122338 GSM5122338 GEO Accession GSM5122338 SRX10202075 GSM5122339 SRP308746 SRS8348555 GSM5122339 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122339 GSM5122339 GEO Accession GSM5122339 SRX10202076 GSM5122340 SRP308746 SRS8348556 GSM5122340 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122340 GSM5122340 GEO Accession GSM5122340 SRX10202077 GSM5122341 SRP308746 SRS8348558 GSM5122341 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122341 GSM5122341 GEO Accession GSM5122341 SRX10202078 GSM5122342 SRP308746 SRS8348557 GSM5122342 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122342 GSM5122342 GEO Accession GSM5122342 SRX10202079 GSM5122343 SRP308746 SRS8348559 GSM5122343 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122343 GSM5122343 GEO Accession GSM5122343 SRX10202080 GSM5122344 SRP308746 SRS8348560 GSM5122344 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122344 GSM5122344 GEO Accession GSM5122344 SRX10202081 GSM5122345 SRP308746 SRS8348561 GSM5122345 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122345 GSM5122345 GEO Accession GSM5122345 SRX10202082 GSM5122346 SRP308746 SRS8348562 GSM5122346 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122346 GSM5122346 GEO Accession GSM5122346 SRX10202083 GSM5122347 SRP308746 SRS8348563 GSM5122347 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122347 GSM5122347 GEO Accession GSM5122347 SRX10202084 GSM5122348 SRP308746 SRS8348564 GSM5122348 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122348 GSM5122348 GEO Accession GSM5122348 SRX10202085 GSM5122349 SRP308746 SRS8348565 GSM5122349 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122349 GSM5122349 GEO Accession GSM5122349 SRX10202086 GSM5122350 SRP308746 SRS8348566 GSM5122350 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122350 GSM5122350 GEO Accession GSM5122350 SRX10202087 GSM5122351 SRP308746 SRS8348567 GSM5122351 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122351 GSM5122351 GEO Accession GSM5122351 SRX10202088 GSM5122352 SRP308746 SRS8348568 GSM5122352 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122352 GSM5122352 GEO Accession GSM5122352 SRX10202089 GSM5122353 SRP308746 SRS8348571 GSM5122353 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122353 GSM5122353 GEO Accession GSM5122353 SRX10202090 GSM5122354 SRP308746 SRS8348569 GSM5122354 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122354 GSM5122354 GEO Accession GSM5122354 SRX10202091 GSM5122355 SRP308746 SRS8348570 GSM5122355 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122355 GSM5122355 GEO Accession GSM5122355 SRX10202092 GSM5122356 SRP308746 SRS8348572 GSM5122356 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122356 GSM5122356 GEO Accession GSM5122356 SRX10202093 GSM5122357 SRP308746 SRS8348573 GSM5122357 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122357 GSM5122357 GEO Accession GSM5122357 SRX10202094 GSM5122358 SRP308746 SRS8348574 GSM5122358 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122358 GSM5122358 GEO Accession GSM5122358 SRX10202095 GSM5122359 SRP308746 SRS8348576 GSM5122359 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122359 GSM5122359 GEO Accession GSM5122359 SRX10202096 GSM5122360 SRP308746 SRS8348575 GSM5122360 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122360 GSM5122360 GEO Accession GSM5122360 SRX10202097 GSM5122361 SRP308746 SRS8348579 GSM5122361 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122361 GSM5122361 GEO Accession GSM5122361 SRX10202098 GSM5122362 SRP308746 SRS8348577 GSM5122362 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122362 GSM5122362 GEO Accession GSM5122362 SRX10202099 GSM5122363 SRP308746 SRS8348578 GSM5122363 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122363 GSM5122363 GEO Accession GSM5122363 SRX10202100 GSM5122364 SRP308746 SRS8348580 GSM5122364 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122364 GSM5122364 GEO Accession GSM5122364 SRX10202101 GSM5122365 SRP308746 SRS8348581 GSM5122365 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122365 GSM5122365 GEO Accession GSM5122365 SRX10202102 GSM5122366 SRP308746 SRS8348582 GSM5122366 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122366 GSM5122366 GEO Accession GSM5122366 SRX10202103 GSM5122367 SRP308746 SRS8348583 GSM5122367 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122367 GSM5122367 GEO Accession GSM5122367 SRX10202104 GSM5122368 SRP308746 SRS8348584 GSM5122368 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122368 GSM5122368 GEO Accession GSM5122368 SRX10202105 GSM5122369 SRP308746 SRS8348585 GSM5122369 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122369 GSM5122369 GEO Accession GSM5122369 SRX10202106 GSM5122370 SRP308746 SRS8348586 GSM5122370 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122370 GSM5122370 GEO Accession GSM5122370 SRX10202107 GSM5122371 SRP308746 SRS8348587 GSM5122371 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122371 GSM5122371 GEO Accession GSM5122371 SRX10202108 GSM5122372 SRP308746 SRS8348588 GSM5122372 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122372 GSM5122372 GEO Accession GSM5122372 SRX10202109 GSM5122373 SRP308746 SRS8348590 GSM5122373 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122373 GSM5122373 GEO Accession GSM5122373 SRX10202110 GSM5122374 SRP308746 SRS8348591 GSM5122374 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122374 GSM5122374 GEO Accession GSM5122374 SRX10202111 GSM5122375 SRP308746 SRS8348589 GSM5122375 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122375 GSM5122375 GEO Accession GSM5122375 SRX10202112 GSM5122376 SRP308746 SRS8348593 GSM5122376 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122376 GSM5122376 GEO Accession GSM5122376 SRX10202113 GSM5122377 SRP308746 SRS8348592 GSM5122377 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122377 GSM5122377 GEO Accession GSM5122377 SRX10202114 GSM5122378 SRP308746 SRS8348594 GSM5122378 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122378 GSM5122378 GEO Accession GSM5122378 SRX10202115 GSM5122379 SRP308746 SRS8348595 GSM5122379 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122379 GSM5122379 GEO Accession GSM5122379 SRX10202116 GSM5122380 SRP308746 SRS8348596 GSM5122380 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122380 GSM5122380 GEO Accession GSM5122380 SRX10202117 GSM5122381 SRP308746 SRS8348597 GSM5122381 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122381 GSM5122381 GEO Accession GSM5122381 SRX10202118 GSM5122382 SRP308746 SRS8348600 GSM5122382 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122382 GSM5122382 GEO Accession GSM5122382 SRX10202119 GSM5122383 SRP308746 SRS8348598 GSM5122383 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122383 GSM5122383 GEO Accession GSM5122383 SRX10202120 GSM5122384 SRP308746 SRS8348599 GSM5122384 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122384 GSM5122384 GEO Accession GSM5122384 SRX10202121 GSM5122385 SRP308746 SRS8348601 GSM5122385 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122385 GSM5122385 GEO Accession GSM5122385 SRX10202122 GSM5122386 SRP308746 SRS8348602 GSM5122386 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122386 GSM5122386 GEO Accession GSM5122386 SRX10202123 GSM5122387 SRP308746 SRS8348603 GSM5122387 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122387 GSM5122387 GEO Accession GSM5122387 SRX10202124 GSM5122388 SRP308746 SRS8348604 GSM5122388 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122388 GSM5122388 GEO Accession GSM5122388 SRX10202125 GSM5122389 SRP308746 SRS8348605 GSM5122389 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122389 GSM5122389 GEO Accession GSM5122389 SRX10202126 GSM5122390 SRP308746 SRS8348606 GSM5122390 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122390 GSM5122390 GEO Accession GSM5122390 SRX10202127 GSM5122391 SRP308746 SRS8348607 GSM5122391 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122391 GSM5122391 GEO Accession GSM5122391 SRX10202128 GSM5122392 SRP308746 SRS8348608 GSM5122392 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122392 GSM5122392 GEO Accession GSM5122392 SRX10202129 GSM5122393 SRP308746 SRS8348609 GSM5122393 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122393 GSM5122393 GEO Accession GSM5122393 SRX10202130 GSM5122394 SRP308746 SRS8348610 GSM5122394 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122394 GSM5122394 GEO Accession GSM5122394 SRX10202131 GSM5122395 SRP308746 SRS8348611 GSM5122395 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122395 GSM5122395 GEO Accession GSM5122395 SRX10202132 GSM5122396 SRP308746 SRS8348612 GSM5122396 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122396 GSM5122396 GEO Accession GSM5122396 SRX10202133 GSM5122397 SRP308746 SRS8348614 GSM5122397 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122397 GSM5122397 GEO Accession GSM5122397 SRX10202134 GSM5122398 SRP308746 SRS8348613 GSM5122398 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122398 GSM5122398 GEO Accession GSM5122398 SRX10202135 GSM5122399 SRP308746 SRS8348615 GSM5122399 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122399 GSM5122399 GEO Accession GSM5122399 SRX10202136 GSM5122400 SRP308746 SRS8348617 GSM5122400 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122400 GSM5122400 GEO Accession GSM5122400 SRX10202137 GSM5122401 SRP308746 SRS8348616 GSM5122401 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122401 GSM5122401 GEO Accession GSM5122401 SRX10202138 GSM5122402 SRP308746 SRS8348618 GSM5122402 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122402 GSM5122402 GEO Accession GSM5122402 SRX10202139 GSM5122403 SRP308746 SRS8348619 GSM5122403 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122403 GSM5122403 GEO Accession GSM5122403 SRX10202140 GSM5122404 SRP308746 SRS8348621 GSM5122404 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122404 GSM5122404 GEO Accession GSM5122404 SRX10202141 GSM5122405 SRP308746 SRS8348620 GSM5122405 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122405 GSM5122405 GEO Accession GSM5122405 SRX10202142 GSM5122406 SRP308746 SRS8348622 GSM5122406 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122406 GSM5122406 GEO Accession GSM5122406 SRX10202143 GSM5122407 SRP308746 SRS8348624 GSM5122407 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122407 GSM5122407 GEO Accession GSM5122407 SRX10202144 GSM5122408 SRP308746 SRS8348623 GSM5122408 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122408 GSM5122408 GEO Accession GSM5122408 SRX10202145 GSM5122409 SRP308746 SRS8348625 GSM5122409 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122409 GSM5122409 GEO Accession GSM5122409 SRX10202146 GSM5122410 SRP308746 SRS8348626 GSM5122410 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122410 GSM5122410 GEO Accession GSM5122410 SRX10202147 GSM5122411 SRP308746 SRS8348627 GSM5122411 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122411 GSM5122411 GEO Accession GSM5122411 SRX10202148 GSM5122412 SRP308746 SRS8348628 GSM5122412 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122412 GSM5122412 GEO Accession GSM5122412 SRX10202149 GSM5122413 SRP308746 SRS8348629 GSM5122413 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122413 GSM5122413 GEO Accession GSM5122413 SRX10202150 GSM5122414 SRP308746 SRS8348631 GSM5122414 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122414 GSM5122414 GEO Accession GSM5122414 SRX10202151 GSM5122415 SRP308746 SRS8348630 GSM5122415 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122415 GSM5122415 GEO Accession GSM5122415 SRX10202152 GSM5122416 SRP308746 SRS8348633 GSM5122416 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122416 GSM5122416 GEO Accession GSM5122416 SRX10202153 GSM5122417 SRP308746 SRS8348632 GSM5122417 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122417 GSM5122417 GEO Accession GSM5122417 SRX10202154 GSM5122418 SRP308746 SRS8348634 GSM5122418 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122418 GSM5122418 GEO Accession GSM5122418 SRX10202155 GSM5122419 SRP308746 SRS8348635 GSM5122419 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122419 GSM5122419 GEO Accession GSM5122419 SRX10202156 GSM5122420 SRP308746 SRS8348636 GSM5122420 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122420 GSM5122420 GEO Accession GSM5122420 SRX10202157 GSM5122421 SRP308746 SRS8348637 GSM5122421 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122421 GSM5122421 GEO Accession GSM5122421 SRX10202158 GSM5122422 SRP308746 SRS8348639 GSM5122422 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122422 GSM5122422 GEO Accession GSM5122422 SRX10202159 GSM5122423 SRP308746 SRS8348638 GSM5122423 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122423 GSM5122423 GEO Accession GSM5122423 SRX10202160 GSM5122424 SRP308746 SRS8348640 GSM5122424 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122424 GSM5122424 GEO Accession GSM5122424 SRX10202161 GSM5122425 SRP308746 SRS8348641 GSM5122425 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122425 GSM5122425 GEO Accession GSM5122425 SRX10202162 GSM5122426 SRP308746 SRS8348642 GSM5122426 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122426 GSM5122426 GEO Accession GSM5122426 SRX10202163 GSM5122427 SRP308746 SRS8348644 GSM5122427 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122427 GSM5122427 GEO Accession GSM5122427 SRX10202164 GSM5122428 SRP308746 SRS8348643 GSM5122428 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122428 GSM5122428 GEO Accession GSM5122428 SRX10202165 GSM5122429 SRP308746 SRS8348645 GSM5122429 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122429 GSM5122429 GEO Accession GSM5122429 SRX10202166 GSM5122430 SRP308746 SRS8348648 GSM5122430 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122430 GSM5122430 GEO Accession GSM5122430 SRX10202167 GSM5122431 SRP308746 SRS8348646 GSM5122431 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122431 GSM5122431 GEO Accession GSM5122431 SRX10202168 GSM5122432 SRP308746 SRS8348647 GSM5122432 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122432 GSM5122432 GEO Accession GSM5122432 SRX10202169 GSM5122433 SRP308746 SRS8348650 GSM5122433 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122433 GSM5122433 GEO Accession GSM5122433 SRX10202170 GSM5122434 SRP308746 SRS8348649 GSM5122434 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122434 GSM5122434 GEO Accession GSM5122434 SRX10202171 GSM5122435 SRP308746 SRS8348651 GSM5122435 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122435 GSM5122435 GEO Accession GSM5122435 SRX10202172 GSM5122436 SRP308746 SRS8348652 GSM5122436 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122436 GSM5122436 GEO Accession GSM5122436 SRX10202173 GSM5122437 SRP308746 SRS8348655 GSM5122437 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122437 GSM5122437 GEO Accession GSM5122437 SRX10202174 GSM5122438 SRP308746 SRS8348654 GSM5122438 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122438 GSM5122438 GEO Accession GSM5122438 SRX10202175 GSM5122439 SRP308746 SRS8348653 GSM5122439 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122439 GSM5122439 GEO Accession GSM5122439 SRX10202176 GSM5122440 SRP308746 SRS8348656 GSM5122440 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122440 GSM5122440 GEO Accession GSM5122440 SRX10202177 GSM5122441 SRP308746 SRS8348657 GSM5122441 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122441 GSM5122441 GEO Accession GSM5122441 SRX10202178 GSM5122442 SRP308746 SRS8348658 GSM5122442 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122442 GSM5122442 GEO Accession GSM5122442 SRX10202179 GSM5122443 SRP308746 SRS8348659 GSM5122443 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122443 GSM5122443 GEO Accession GSM5122443 SRX10202180 GSM5122444 SRP308746 SRS8348661 GSM5122444 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122444 GSM5122444 GEO Accession GSM5122444 SRX10202181 GSM5122445 SRP308746 SRS8348660 GSM5122445 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122445 GSM5122445 GEO Accession GSM5122445 SRX10202182 GSM5122446 SRP308746 SRS8348662 GSM5122446 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122446 GSM5122446 GEO Accession GSM5122446 SRX10202183 GSM5122447 SRP308746 SRS8348663 GSM5122447 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122447 GSM5122447 GEO Accession GSM5122447 SRX10202184 GSM5122448 SRP308746 SRS8348665 GSM5122448 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122448 GSM5122448 GEO Accession GSM5122448 SRX10202185 GSM5122449 SRP308746 SRS8348664 GSM5122449 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122449 GSM5122449 GEO Accession GSM5122449 SRX10202186 GSM5122450 SRP308746 SRS8348666 GSM5122450 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122450 GSM5122450 GEO Accession GSM5122450 SRX10202187 GSM5122451 SRP308746 SRS8348667 GSM5122451 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122451 GSM5122451 GEO Accession GSM5122451 SRX10202188 GSM5122452 SRP308746 SRS8348670 GSM5122452 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122452 GSM5122452 GEO Accession GSM5122452 SRX10202189 GSM5122453 SRP308746 SRS8348668 GSM5122453 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122453 GSM5122453 GEO Accession GSM5122453 SRX10202190 GSM5122454 SRP308746 SRS8348669 GSM5122454 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122454 GSM5122454 GEO Accession GSM5122454 SRX10202191 GSM5122455 SRP308746 SRS8348673 GSM5122455 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122455 GSM5122455 GEO Accession GSM5122455 SRX10202192 GSM5122456 SRP308746 SRS8348672 GSM5122456 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122456 GSM5122456 GEO Accession GSM5122456 SRX10202193 GSM5122457 SRP308746 SRS8348671 GSM5122457 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122457 GSM5122457 GEO Accession GSM5122457 SRX10202194 GSM5122458 SRP308746 SRS8348674 GSM5122458 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122458 GSM5122458 GEO Accession GSM5122458 SRX10202195 GSM5122459 SRP308746 SRS8348675 GSM5122459 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122459 GSM5122459 GEO Accession GSM5122459 SRX10202196 GSM5122460 SRP308746 SRS8348677 GSM5122460 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122460 GSM5122460 GEO Accession GSM5122460 SRX10202197 GSM5122461 SRP308746 SRS8348676 GSM5122461 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122461 GSM5122461 GEO Accession GSM5122461 SRX10202198 GSM5122462 SRP308746 SRS8348678 GSM5122462 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122462 GSM5122462 GEO Accession GSM5122462 SRX10202199 GSM5122463 SRP308746 SRS8348680 GSM5122463 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122463 GSM5122463 GEO Accession GSM5122463 SRX10202200 GSM5122464 SRP308746 SRS8348679 GSM5122464 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122464 GSM5122464 GEO Accession GSM5122464 SRX10202201 GSM5122465 SRP308746 SRS8348682 GSM5122465 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122465 GSM5122465 GEO Accession GSM5122465 SRX10202202 GSM5122466 SRP308746 SRS8348681 GSM5122466 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122466 GSM5122466 GEO Accession GSM5122466 SRX10202203 GSM5122467 SRP308746 SRS8348684 GSM5122467 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122467 GSM5122467 GEO Accession GSM5122467 SRX10202204 GSM5122468 SRP308746 SRS8348683 GSM5122468 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122468 GSM5122468 GEO Accession GSM5122468 SRX10202205 GSM5122469 SRP308746 SRS8348687 GSM5122469 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122469 GSM5122469 GEO Accession GSM5122469 SRX10202206 GSM5122470 SRP308746 SRS8348686 GSM5122470 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122470 GSM5122470 GEO Accession GSM5122470 SRX10202207 GSM5122471 SRP308746 SRS8348685 GSM5122471 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122471 GSM5122471 GEO Accession GSM5122471 SRX10202208 GSM5122472 SRP308746 SRS8348690 GSM5122472 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122472 GSM5122472 GEO Accession GSM5122472 SRX10202209 GSM5122473 SRP308746 SRS8348689 GSM5122473 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122473 GSM5122473 GEO Accession GSM5122473 SRX10202210 GSM5122474 SRP308746 SRS8348688 GSM5122474 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122474 GSM5122474 GEO Accession GSM5122474 SRX10202211 GSM5122475 SRP308746 SRS8348691 GSM5122475 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122475 GSM5122475 GEO Accession GSM5122475 SRX10202212 GSM5122476 SRP308746 SRS8348692 GSM5122476 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122476 GSM5122476 GEO Accession GSM5122476 SRX10202213 GSM5122477 SRP308746 SRS8348693 GSM5122477 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122477 GSM5122477 GEO Accession GSM5122477 SRX10202214 GSM5122478 SRP308746 SRS8348694 GSM5122478 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122478 GSM5122478 GEO Accession GSM5122478 SRX10202215 GSM5122479 SRP308746 SRS8348695 GSM5122479 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122479 GSM5122479 GEO Accession GSM5122479 SRX10202216 GSM5122480 SRP308746 SRS8348696 GSM5122480 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122480 GSM5122480 GEO Accession GSM5122480 SRX10202217 GSM5122481 SRP308746 SRS8348697 GSM5122481 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122481 GSM5122481 GEO Accession GSM5122481 SRX10202218 GSM5122482 SRP308746 SRS8348698 GSM5122482 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122482 GSM5122482 GEO Accession GSM5122482 SRX10202219 GSM5122483 SRP308746 SRS8348699 GSM5122483 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122483 GSM5122483 GEO Accession GSM5122483 SRX10202220 GSM5122484 SRP308746 SRS8348700 GSM5122484 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122484 GSM5122484 GEO Accession GSM5122484 SRX10202221 GSM5122485 SRP308746 SRS8348701 GSM5122485 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122485 GSM5122485 GEO Accession GSM5122485 SRX10202222 GSM5122486 SRP308746 SRS8348702 GSM5122486 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122486 GSM5122486 GEO Accession GSM5122486 SRX10202223 GSM5122487 SRP308746 SRS8348703 GSM5122487 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122487 GSM5122487 GEO Accession GSM5122487 SRX10202224 GSM5122488 SRP308746 SRS8348704 GSM5122488 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122488 GSM5122488 GEO Accession GSM5122488 SRX10202225 GSM5122489 SRP308746 SRS8348706 GSM5122489 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122489 GSM5122489 GEO Accession GSM5122489 SRX10202226 GSM5122490 SRP308746 SRS8348705 GSM5122490 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122490 GSM5122490 GEO Accession GSM5122490 SRX10202227 GSM5122491 SRP308746 SRS8348707 GSM5122491 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122491 GSM5122491 GEO Accession GSM5122491 SRX10202228 GSM5122492 SRP308746 SRS8348708 GSM5122492 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122492 GSM5122492 GEO Accession GSM5122492 SRX10202229 GSM5122493 SRP308746 SRS8348710 GSM5122493 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122493 GSM5122493 GEO Accession GSM5122493 SRX10202230 GSM5122494 SRP308746 SRS8348709 GSM5122494 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122494 GSM5122494 GEO Accession GSM5122494 SRX10202231 GSM5122495 SRP308746 SRS8348711 GSM5122495 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122495 GSM5122495 GEO Accession GSM5122495 SRX10202232 GSM5122496 SRP308746 SRS8348712 GSM5122496 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122496 GSM5122496 GEO Accession GSM5122496 SRX10202233 GSM5122497 SRP308746 SRS8348713 GSM5122497 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122497 GSM5122497 GEO Accession GSM5122497 SRX10202234 GSM5122498 SRP308746 SRS8348714 GSM5122498 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122498 GSM5122498 GEO Accession GSM5122498 SRX10202235 GSM5122499 SRP308746 SRS8348715 GSM5122499 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122499 GSM5122499 GEO Accession GSM5122499 SRX10202236 GSM5122500 SRP308746 SRS8348716 GSM5122500 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122500 GSM5122500 GEO Accession GSM5122500 SRX10202237 GSM5122501 SRP308746 SRS8348718 GSM5122501 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122501 GSM5122501 GEO Accession GSM5122501 SRX10202238 GSM5122502 SRP308746 SRS8348717 GSM5122502 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122502 GSM5122502 GEO Accession GSM5122502 SRX10202239 GSM5122503 SRP308746 SRS8348719 GSM5122503 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122503 GSM5122503 GEO Accession GSM5122503 SRX10202240 GSM5122504 SRP308746 SRS8348720 GSM5122504 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122504 GSM5122504 GEO Accession GSM5122504 SRX10202241 GSM5122505 SRP308746 SRS8348721 GSM5122505 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122505 GSM5122505 GEO Accession GSM5122505 SRX10202242 GSM5122506 SRP308746 SRS8348722 GSM5122506 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122506 GSM5122506 GEO Accession GSM5122506 SRX10202243 GSM5122507 SRP308746 SRS8348725 GSM5122507 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122507 GSM5122507 GEO Accession GSM5122507 SRX10202244 GSM5122508 SRP308746 SRS8348723 GSM5122508 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122508 GSM5122508 GEO Accession GSM5122508 SRX10202245 GSM5122509 SRP308746 SRS8348724 GSM5122509 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122509 GSM5122509 GEO Accession GSM5122509 SRX10202246 GSM5122510 SRP308746 SRS8348726 GSM5122510 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122510 GSM5122510 GEO Accession GSM5122510 SRX10202247 GSM5122511 SRP308746 SRS8348729 GSM5122511 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122511 GSM5122511 GEO Accession GSM5122511 SRX10202248 GSM5122512 SRP308746 SRS8348728 GSM5122512 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122512 GSM5122512 GEO Accession GSM5122512 SRX10202249 GSM5122513 SRP308746 SRS8348727 GSM5122513 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122513 GSM5122513 GEO Accession GSM5122513 SRX10202250 GSM5122514 SRP308746 SRS8348730 GSM5122514 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122514 GSM5122514 GEO Accession GSM5122514 SRX10202251 GSM5122515 SRP308746 SRS8348731 GSM5122515 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122515 GSM5122515 GEO Accession GSM5122515 SRX10202252 GSM5122516 SRP308746 SRS8348733 GSM5122516 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122516 GSM5122516 GEO Accession GSM5122516 SRX10202253 GSM5122517 SRP308746 SRS8348732 GSM5122517 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122517 GSM5122517 GEO Accession GSM5122517 SRX10202254 GSM5122518 SRP308746 SRS8348735 GSM5122518 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122518 GSM5122518 GEO Accession GSM5122518 SRX10202255 GSM5122519 SRP308746 SRS8348734 GSM5122519 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122519 GSM5122519 GEO Accession GSM5122519 SRX10202256 GSM5122520 SRP308746 SRS8348736 GSM5122520 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122520 GSM5122520 GEO Accession GSM5122520 SRX10202257 GSM5122521 SRP308746 SRS8348737 GSM5122521 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122521 GSM5122521 GEO Accession GSM5122521 SRX10202258 GSM5122522 SRP308746 SRS8348738 GSM5122522 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122522 GSM5122522 GEO Accession GSM5122522 SRX10202259 GSM5122523 SRP308746 SRS8348739 GSM5122523 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122523 GSM5122523 GEO Accession GSM5122523 SRX10202260 GSM5122524 SRP308746 SRS8348740 GSM5122524 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122524 GSM5122524 GEO Accession GSM5122524 SRX10202261 GSM5122525 SRP308746 SRS8348741 GSM5122525 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122525 GSM5122525 GEO Accession GSM5122525 SRX10202262 GSM5122526 SRP308746 SRS8348742 GSM5122526 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122526 GSM5122526 GEO Accession GSM5122526 SRX10202263 GSM5122527 SRP308746 SRS8348744 GSM5122527 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122527 GSM5122527 GEO Accession GSM5122527 SRX10202264 GSM5122528 SRP308746 SRS8348743 GSM5122528 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122528 GSM5122528 GEO Accession GSM5122528 SRX10202265 GSM5122529 SRP308746 SRS8348745 GSM5122529 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122529 GSM5122529 GEO Accession GSM5122529 SRX10202266 GSM5122530 SRP308746 SRS8348746 GSM5122530 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122530 GSM5122530 GEO Accession GSM5122530 SRX10202267 GSM5122531 SRP308746 SRS8348747 GSM5122531 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122531 GSM5122531 GEO Accession GSM5122531 SRX10202268 GSM5122532 SRP308746 SRS8348748 GSM5122532 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122532 GSM5122532 GEO Accession GSM5122532 SRX10202269 GSM5122533 SRP308746 SRS8348749 GSM5122533 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122533 GSM5122533 GEO Accession GSM5122533 SRX10202270 GSM5122534 SRP308746 SRS8348750 GSM5122534 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122534 GSM5122534 GEO Accession GSM5122534 SRX10202271 GSM5122535 SRP308746 SRS8348751 GSM5122535 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122535 GSM5122535 GEO Accession GSM5122535 SRX10202272 GSM5122536 SRP308746 SRS8348752 GSM5122536 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122536 GSM5122536 GEO Accession GSM5122536 SRX10202273 GSM5122537 SRP308746 SRS8348755 GSM5122537 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122537 GSM5122537 GEO Accession GSM5122537 SRX10202274 GSM5122538 SRP308746 SRS8348754 GSM5122538 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122538 GSM5122538 GEO Accession GSM5122538 SRX10202275 GSM5122539 SRP308746 SRS8348753 GSM5122539 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122539 GSM5122539 GEO Accession GSM5122539 SRX10202276 GSM5122540 SRP308746 SRS8348757 GSM5122540 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122540 GSM5122540 GEO Accession GSM5122540 SRX10202277 GSM5122541 SRP308746 SRS8348756 GSM5122541 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122541 GSM5122541 GEO Accession GSM5122541 SRX10202278 GSM5122542 SRP308746 SRS8348758 GSM5122542 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122542 GSM5122542 GEO Accession GSM5122542 SRX10202279 GSM5122543 SRP308746 SRS8348761 GSM5122543 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122543 GSM5122543 GEO Accession GSM5122543 SRX10202280 GSM5122544 SRP308746 SRS8348759 GSM5122544 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122544 GSM5122544 GEO Accession GSM5122544 SRX10202281 GSM5122545 SRP308746 SRS8348760 GSM5122545 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122545 GSM5122545 GEO Accession GSM5122545 SRX10202282 GSM5122546 SRP308746 SRS8348763 GSM5122546 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122546 GSM5122546 GEO Accession GSM5122546 SRX10202283 GSM5122547 SRP308746 SRS8348764 GSM5122547 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122547 GSM5122547 GEO Accession GSM5122547 SRX10202284 GSM5122548 SRP308746 SRS8348762 GSM5122548 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122548 GSM5122548 GEO Accession GSM5122548 SRX10202285 GSM5122549 SRP308746 SRS8348765 GSM5122549 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122549 GSM5122549 GEO Accession GSM5122549 SRX10202286 GSM5122550 SRP308746 SRS8348766 GSM5122550 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122550 GSM5122550 GEO Accession GSM5122550 SRX10202287 GSM5122551 SRP308746 SRS8348767 GSM5122551 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122551 GSM5122551 GEO Accession GSM5122551 SRX10202288 GSM5122552 SRP308746 SRS8348768 GSM5122552 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122552 GSM5122552 GEO Accession GSM5122552 SRX10202289 GSM5122553 SRP308746 SRS8348769 GSM5122553 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122553 GSM5122553 GEO Accession GSM5122553 SRX10202290 GSM5122554 SRP308746 SRS8348770 GSM5122554 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122554 GSM5122554 GEO Accession GSM5122554 SRX10202291 GSM5122555 SRP308746 SRS8348773 GSM5122555 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122555 GSM5122555 GEO Accession GSM5122555 SRX10202292 GSM5122556 SRP308746 SRS8348771 GSM5122556 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122556 GSM5122556 GEO Accession GSM5122556 SRX10202293 GSM5122557 SRP308746 SRS8348772 GSM5122557 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122557 GSM5122557 GEO Accession GSM5122557 SRX10202294 GSM5122558 SRP308746 SRS8348774 GSM5122558 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122558 GSM5122558 GEO Accession GSM5122558 SRX10202295 GSM5122559 SRP308746 SRS8348775 GSM5122559 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122559 GSM5122559 GEO Accession GSM5122559 SRX10202296 GSM5122560 SRP308746 SRS8348776 GSM5122560 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122560 GSM5122560 GEO Accession GSM5122560 SRX10202297 GSM5122561 SRP308746 SRS8348777 GSM5122561 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122561 GSM5122561 GEO Accession GSM5122561 SRX10202298 GSM5122562 SRP308746 SRS8348781 GSM5122562 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122562 GSM5122562 GEO Accession GSM5122562 SRX10202299 GSM5122563 SRP308746 SRS8348778 GSM5122563 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122563 GSM5122563 GEO Accession GSM5122563 SRX10202300 GSM5122564 SRP308746 SRS8348779 GSM5122564 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122564 GSM5122564 GEO Accession GSM5122564 SRX10202301 GSM5122565 SRP308746 SRS8348780 GSM5122565 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122565 GSM5122565 GEO Accession GSM5122565 SRX10202302 GSM5122566 SRP308746 SRS8348782 GSM5122566 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122566 GSM5122566 GEO Accession GSM5122566 SRX10202303 GSM5122567 SRP308746 SRS8348783 GSM5122567 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122567 GSM5122567 GEO Accession GSM5122567 SRX10202304 GSM5122568 SRP308746 SRS8348786 GSM5122568 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122568 GSM5122568 GEO Accession GSM5122568 SRX10202305 GSM5122569 SRP308746 SRS8348784 GSM5122569 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122569 GSM5122569 GEO Accession GSM5122569 SRX10202306 GSM5122570 SRP308746 SRS8348785 GSM5122570 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122570 GSM5122570 GEO Accession GSM5122570 SRX10202307 GSM5122571 SRP308746 SRS8348787 GSM5122571 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122571 GSM5122571 GEO Accession GSM5122571 SRX10202308 GSM5122572 SRP308746 SRS8348788 GSM5122572 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122572 GSM5122572 GEO Accession GSM5122572 SRX10202309 GSM5122573 SRP308746 SRS8348789 GSM5122573 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122573 GSM5122573 GEO Accession GSM5122573 SRX10202310 GSM5122574 SRP308746 SRS8348790 GSM5122574 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122574 GSM5122574 GEO Accession GSM5122574 SRX10202311 GSM5122575 SRP308746 SRS8348791 GSM5122575 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122575 GSM5122575 GEO Accession GSM5122575 SRX10202312 GSM5122576 SRP308746 SRS8348792 GSM5122576 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122576 GSM5122576 GEO Accession GSM5122576 SRX10202313 GSM5122577 SRP308746 SRS8348794 GSM5122577 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122577 GSM5122577 GEO Accession GSM5122577 SRX10202314 GSM5122578 SRP308746 SRS8348793 GSM5122578 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122578 GSM5122578 GEO Accession GSM5122578 SRX10202315 GSM5122579 SRP308746 SRS8348795 GSM5122579 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122579 GSM5122579 GEO Accession GSM5122579 SRX10202316 GSM5122580 SRP308746 SRS8348797 GSM5122580 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122580 GSM5122580 GEO Accession GSM5122580 SRX10202317 GSM5122581 SRP308746 SRS8348796 GSM5122581 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122581 GSM5122581 GEO Accession GSM5122581 SRX10202318 GSM5122582 SRP308746 SRS8348798 GSM5122582 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122582 GSM5122582 GEO Accession GSM5122582 SRX10202319 GSM5122583 SRP308746 SRS8348799 GSM5122583 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122583 GSM5122583 GEO Accession GSM5122583 SRX10202320 GSM5122584 SRP308746 SRS8348800 GSM5122584 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122584 GSM5122584 GEO Accession GSM5122584 SRX10202321 GSM5122585 SRP308746 SRS8348801 GSM5122585 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122585 GSM5122585 GEO Accession GSM5122585 SRX10202322 GSM5122586 SRP308746 SRS8348802 GSM5122586 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122586 GSM5122586 GEO Accession GSM5122586 SRX10202323 GSM5122587 SRP308746 SRS8348803 GSM5122587 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122587 GSM5122587 GEO Accession GSM5122587 SRX10202324 GSM5122588 SRP308746 SRS8348805 GSM5122588 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122588 GSM5122588 GEO Accession GSM5122588 SRX10202325 GSM5122589 SRP308746 SRS8348804 GSM5122589 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122589 GSM5122589 GEO Accession GSM5122589 SRX10202326 GSM5122590 SRP308746 SRS8348806 GSM5122590 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122590 GSM5122590 GEO Accession GSM5122590 SRX10202327 GSM5122591 SRP308746 SRS8348808 GSM5122591 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122591 GSM5122591 GEO Accession GSM5122591 SRX10202328 GSM5122592 SRP308746 SRS8348807 GSM5122592 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122592 GSM5122592 GEO Accession GSM5122592 SRX10202329 GSM5122593 SRP308746 SRS8348809 GSM5122593 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122593 GSM5122593 GEO Accession GSM5122593 SRX10202330 GSM5122594 SRP308746 SRS8348811 GSM5122594 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122594 GSM5122594 GEO Accession GSM5122594 SRX10202331 GSM5122595 SRP308746 SRS8348810 GSM5122595 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122595 GSM5122595 GEO Accession GSM5122595 SRX10202332 GSM5122596 SRP308746 SRS8348813 GSM5122596 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122596 GSM5122596 GEO Accession GSM5122596 SRX10202333 GSM5122597 SRP308746 SRS8348812 GSM5122597 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122597 GSM5122597 GEO Accession GSM5122597 SRX10202334 GSM5122598 SRP308746 SRS8348814 GSM5122598 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122598 GSM5122598 GEO Accession GSM5122598 SRX10202335 GSM5122599 SRP308746 SRS8348815 GSM5122599 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122599 GSM5122599 GEO Accession GSM5122599 SRX10202336 GSM5122600 SRP308746 SRS8348817 GSM5122600 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122600 GSM5122600 GEO Accession GSM5122600 SRX10202337 GSM5122601 SRP308746 SRS8348819 GSM5122601 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122601 GSM5122601 GEO Accession GSM5122601 SRX10202338 GSM5122602 SRP308746 SRS8348816 GSM5122602 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122602 GSM5122602 GEO Accession GSM5122602 SRX10202339 GSM5122603 SRP308746 SRS8348818 GSM5122603 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122603 GSM5122603 GEO Accession GSM5122603 SRX10202340 GSM5122604 SRP308746 SRS8348820 GSM5122604 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122604 GSM5122604 GEO Accession GSM5122604 SRX10202341 GSM5122605 SRP308746 SRS8348822 GSM5122605 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122605 GSM5122605 GEO Accession GSM5122605 SRX10202342 GSM5122606 SRP308746 SRS8348821 GSM5122606 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122606 GSM5122606 GEO Accession GSM5122606 SRX10202343 GSM5122607 SRP308746 SRS8348823 GSM5122607 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122607 GSM5122607 GEO Accession GSM5122607 SRX10202344 GSM5122608 SRP308746 SRS8348824 GSM5122608 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122608 GSM5122608 GEO Accession GSM5122608 SRX10202345 GSM5122609 SRP308746 SRS8348825 GSM5122609 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122609 GSM5122609 GEO Accession GSM5122609 SRX10202346 GSM5122610 SRP308746 SRS8348826 GSM5122610 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122610 GSM5122610 GEO Accession GSM5122610 SRX10202347 GSM5122611 SRP308746 SRS8348827 GSM5122611 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122611 GSM5122611 GEO Accession GSM5122611 SRX10202348 GSM5122612 SRP308746 SRS8348829 GSM5122612 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122612 GSM5122612 GEO Accession GSM5122612 SRX10202349 GSM5122613 SRP308746 SRS8348830 GSM5122613 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122613 GSM5122613 GEO Accession GSM5122613 SRX10202350 GSM5122614 SRP308746 SRS8348831 GSM5122614 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122614 GSM5122614 GEO Accession GSM5122614 SRX10202351 GSM5122615 SRP308746 SRS8348828 GSM5122615 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122615 GSM5122615 GEO Accession GSM5122615 SRX10202352 GSM5122616 SRP308746 SRS8348832 GSM5122616 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122616 GSM5122616 GEO Accession GSM5122616 SRX10202353 GSM5122617 SRP308746 SRS8348833 GSM5122617 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122617 GSM5122617 GEO Accession GSM5122617 SRX10202354 GSM5122618 SRP308746 SRS8348834 GSM5122618 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122618 GSM5122618 GEO Accession GSM5122618 SRX10202355 GSM5122619 SRP308746 SRS8348835 GSM5122619 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122619 GSM5122619 GEO Accession GSM5122619 SRX10202356 GSM5122620 SRP308746 SRS8348837 GSM5122620 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122620 GSM5122620 GEO Accession GSM5122620 SRX10202357 GSM5122621 SRP308746 SRS8348836 GSM5122621 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122621 GSM5122621 GEO Accession GSM5122621 SRX10202358 GSM5122622 SRP308746 SRS8348838 GSM5122622 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122622 GSM5122622 GEO Accession GSM5122622 SRX10202359 GSM5122623 SRP308746 SRS8348840 GSM5122623 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122623 GSM5122623 GEO Accession GSM5122623 SRX10202360 GSM5122624 SRP308746 SRS8348839 GSM5122624 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122624 GSM5122624 GEO Accession GSM5122624 SRX10202361 GSM5122625 SRP308746 SRS8348842 GSM5122625 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122625 GSM5122625 GEO Accession GSM5122625 SRX10202362 GSM5122626 SRP308746 SRS8348841 GSM5122626 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122626 GSM5122626 GEO Accession GSM5122626 SRX10202363 GSM5122627 SRP308746 SRS8348843 GSM5122627 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122627 GSM5122627 GEO Accession GSM5122627 SRX10202364 GSM5122628 SRP308746 SRS8348844 GSM5122628 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122628 GSM5122628 GEO Accession GSM5122628 SRX10202365 GSM5122629 SRP308746 SRS8348846 GSM5122629 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122629 GSM5122629 GEO Accession GSM5122629 SRX10202366 GSM5122630 SRP308746 SRS8348845 GSM5122630 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122630 GSM5122630 GEO Accession GSM5122630 SRX10202367 GSM5122631 SRP308746 SRS8348850 GSM5122631 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122631 GSM5122631 GEO Accession GSM5122631 SRX10202368 GSM5122632 SRP308746 SRS8348847 GSM5122632 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122632 GSM5122632 GEO Accession GSM5122632 SRX10202369 GSM5122633 SRP308746 SRS8348849 GSM5122633 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122633 GSM5122633 GEO Accession GSM5122633 SRX10202370 GSM5122634 SRP308746 SRS8348848 GSM5122634 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122634 GSM5122634 GEO Accession GSM5122634 SRX10202371 GSM5122635 SRP308746 SRS8348851 GSM5122635 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122635 GSM5122635 GEO Accession GSM5122635 SRX10202372 GSM5122636 SRP308746 SRS8348852 GSM5122636 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122636 GSM5122636 GEO Accession GSM5122636 SRX10202373 GSM5122637 SRP308746 SRS8348854 GSM5122637 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122637 GSM5122637 GEO Accession GSM5122637 SRX10202374 GSM5122638 SRP308746 SRS8348853 GSM5122638 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122638 GSM5122638 GEO Accession GSM5122638 SRX10202375 GSM5122639 SRP308746 SRS8348855 GSM5122639 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122639 GSM5122639 GEO Accession GSM5122639 SRX10202376 GSM5122640 SRP308746 SRS8348856 GSM5122640 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122640 GSM5122640 GEO Accession GSM5122640 SRX10202377 GSM5122641 SRP308746 SRS8348857 GSM5122641 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122641 GSM5122641 GEO Accession GSM5122641 SRX10202378 GSM5122642 SRP308746 SRS8348858 GSM5122642 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122642 GSM5122642 GEO Accession GSM5122642 SRX10202379 GSM5122643 SRP308746 SRS8348859 GSM5122643 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122643 GSM5122643 GEO Accession GSM5122643 SRX10202380 GSM5122644 SRP308746 SRS8348860 GSM5122644 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122644 GSM5122644 GEO Accession GSM5122644 SRX10202381 GSM5122645 SRP308746 SRS8348861 GSM5122645 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122645 GSM5122645 GEO Accession GSM5122645 SRX10202382 GSM5122646 SRP308746 SRS8348862 GSM5122646 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122646 GSM5122646 GEO Accession GSM5122646 SRX10202383 GSM5122647 SRP308746 SRS8348863 GSM5122647 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122647 GSM5122647 GEO Accession GSM5122647 SRX10202384 GSM5122648 SRP308746 SRS8348864 GSM5122648 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122648 GSM5122648 GEO Accession GSM5122648 SRX10202385 GSM5122649 SRP308746 SRS8348865 GSM5122649 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122649 GSM5122649 GEO Accession GSM5122649 SRX10202386 GSM5122650 SRP308746 SRS8348866 GSM5122650 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122650 GSM5122650 GEO Accession GSM5122650 SRX10202387 GSM5122651 SRP308746 SRS8348868 GSM5122651 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122651 GSM5122651 GEO Accession GSM5122651 SRX10202388 GSM5122652 SRP308746 SRS8348867 GSM5122652 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122652 GSM5122652 GEO Accession GSM5122652 SRX10202389 GSM5122653 SRP308746 SRS8348869 GSM5122653 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122653 GSM5122653 GEO Accession GSM5122653 SRX10202390 GSM5122654 SRP308746 SRS8348870 GSM5122654 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122654 GSM5122654 GEO Accession GSM5122654 SRX10202391 GSM5122655 SRP308746 SRS8348871 GSM5122655 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122655 GSM5122655 GEO Accession GSM5122655 SRX10202392 GSM5122656 SRP308746 SRS8348872 GSM5122656 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122656 GSM5122656 GEO Accession GSM5122656 SRX10202393 GSM5122657 SRP308746 SRS8348873 GSM5122657 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122657 GSM5122657 GEO Accession GSM5122657 SRX10202394 GSM5122658 SRP308746 SRS8348874 GSM5122658 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122658 GSM5122658 GEO Accession GSM5122658 SRX10202395 GSM5122659 SRP308746 SRS8348875 GSM5122659 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122659 GSM5122659 GEO Accession GSM5122659 SRX10202396 GSM5122660 SRP308746 SRS8348876 GSM5122660 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122660 GSM5122660 GEO Accession GSM5122660 SRX10202397 GSM5122661 SRP308746 SRS8348877 GSM5122661 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122661 GSM5122661 GEO Accession GSM5122661 SRX10202398 GSM5122662 SRP308746 SRS8348878 GSM5122662 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122662 GSM5122662 GEO Accession GSM5122662 SRX10202399 GSM5122663 SRP308746 SRS8348879 GSM5122663 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122663 GSM5122663 GEO Accession GSM5122663 SRX10202400 GSM5122664 SRP308746 SRS8348883 GSM5122664 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122664 GSM5122664 GEO Accession GSM5122664 SRX10202401 GSM5122665 SRP308746 SRS8348882 GSM5122665 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122665 GSM5122665 GEO Accession GSM5122665 SRX10202402 GSM5122666 SRP308746 SRS8348880 GSM5122666 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122666 GSM5122666 GEO Accession GSM5122666 SRX10202403 GSM5122667 SRP308746 SRS8348881 GSM5122667 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122667 GSM5122667 GEO Accession GSM5122667 SRX10202404 GSM5122668 SRP308746 SRS8348885 GSM5122668 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122668 GSM5122668 GEO Accession GSM5122668 SRX10202405 GSM5122669 SRP308746 SRS8348884 GSM5122669 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122669 GSM5122669 GEO Accession GSM5122669 SRX10202406 GSM5122670 SRP308746 SRS8348886 GSM5122670 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122670 GSM5122670 GEO Accession GSM5122670 SRX10202407 GSM5122671 SRP308746 SRS8348887 GSM5122671 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122671 GSM5122671 GEO Accession GSM5122671 SRX10202408 GSM5122672 SRP308746 SRS8348888 GSM5122672 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122672 GSM5122672 GEO Accession GSM5122672 SRX10202409 GSM5122673 SRP308746 SRS8348890 GSM5122673 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122673 GSM5122673 GEO Accession GSM5122673 SRX10202410 GSM5122674 SRP308746 SRS8348889 GSM5122674 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122674 GSM5122674 GEO Accession GSM5122674 SRX10202411 GSM5122675 SRP308746 SRS8348891 GSM5122675 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122675 GSM5122675 GEO Accession GSM5122675 SRX10202412 GSM5122676 SRP308746 SRS8348892 GSM5122676 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122676 GSM5122676 GEO Accession GSM5122676 SRX10202413 GSM5122677 SRP308746 SRS8348893 GSM5122677 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122677 GSM5122677 GEO Accession GSM5122677 SRX10202414 GSM5122678 SRP308746 SRS8348896 GSM5122678 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122678 GSM5122678 GEO Accession GSM5122678 SRX10202415 GSM5122679 SRP308746 SRS8348895 GSM5122679 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122679 GSM5122679 GEO Accession GSM5122679 SRX10202416 GSM5122680 SRP308746 SRS8348894 GSM5122680 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122680 GSM5122680 GEO Accession GSM5122680 SRX10202417 GSM5122681 SRP308746 SRS8348897 GSM5122681 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122681 GSM5122681 GEO Accession GSM5122681 SRX10202418 GSM5122682 SRP308746 SRS8348898 GSM5122682 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122682 GSM5122682 GEO Accession GSM5122682 SRX10202419 GSM5122683 SRP308746 SRS8348899 GSM5122683 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122683 GSM5122683 GEO Accession GSM5122683 SRX10202420 GSM5122684 SRP308746 SRS8348900 GSM5122684 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122684 GSM5122684 GEO Accession GSM5122684 SRX10202421 GSM5122685 SRP308746 SRS8348902 GSM5122685 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122685 GSM5122685 GEO Accession GSM5122685 SRX10202422 GSM5122686 SRP308746 SRS8348901 GSM5122686 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122686 GSM5122686 GEO Accession GSM5122686 SRX10202423 GSM5122687 SRP308746 SRS8348903 GSM5122687 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122687 GSM5122687 GEO Accession GSM5122687 SRX10202424 GSM5122688 SRP308746 SRS8348904 GSM5122688 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122688 GSM5122688 GEO Accession GSM5122688 SRX10202425 GSM5122689 SRP308746 SRS8348905 GSM5122689 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122689 GSM5122689 GEO Accession GSM5122689 SRX10202426 GSM5122690 SRP308746 SRS8348906 GSM5122690 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122690 GSM5122690 GEO Accession GSM5122690 SRX10202427 GSM5122691 SRP308746 SRS8348908 GSM5122691 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122691 GSM5122691 GEO Accession GSM5122691 SRX10202428 GSM5122692 SRP308746 SRS8348907 GSM5122692 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122692 GSM5122692 GEO Accession GSM5122692 SRX10202429 GSM5122693 SRP308746 SRS8348909 GSM5122693 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122693 GSM5122693 GEO Accession GSM5122693 SRX10202430 GSM5122694 SRP308746 SRS8348910 GSM5122694 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122694 GSM5122694 GEO Accession GSM5122694 SRX10202431 GSM5122695 SRP308746 SRS8348911 GSM5122695 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122695 GSM5122695 GEO Accession GSM5122695 SRX10202432 GSM5122696 SRP308746 SRS8348912 GSM5122696 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122696 GSM5122696 GEO Accession GSM5122696 SRX10202433 GSM5122697 SRP308746 SRS8348913 GSM5122697 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122697 GSM5122697 GEO Accession GSM5122697 SRX10202434 GSM5122698 SRP308746 SRS8348914 GSM5122698 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122698 GSM5122698 GEO Accession GSM5122698 SRX10202435 GSM5122699 SRP308746 SRS8348915 GSM5122699 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122699 GSM5122699 GEO Accession GSM5122699 SRX10202436 GSM5122700 SRP308746 SRS8348917 GSM5122700 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122700 GSM5122700 GEO Accession GSM5122700 SRX10202437 GSM5122701 SRP308746 SRS8348916 GSM5122701 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122701 GSM5122701 GEO Accession GSM5122701 SRX10202438 GSM5122702 SRP308746 SRS8348918 GSM5122702 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122702 GSM5122702 GEO Accession GSM5122702 SRX10202439 GSM5122703 SRP308746 SRS8348919 GSM5122703 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122703 GSM5122703 GEO Accession GSM5122703 SRX10202440 GSM5122704 SRP308746 SRS8348920 GSM5122704 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122704 GSM5122704 GEO Accession GSM5122704 SRX10202441 GSM5122705 SRP308746 SRS8348921 GSM5122705 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122705 GSM5122705 GEO Accession GSM5122705 SRX10202442 GSM5122706 SRP308746 SRS8348922 GSM5122706 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122706 GSM5122706 GEO Accession GSM5122706 SRX10202443 GSM5122707 SRP308746 SRS8348923 GSM5122707 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122707 GSM5122707 GEO Accession GSM5122707 SRX10202444 GSM5122708 SRP308746 SRS8348924 GSM5122708 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122708 GSM5122708 GEO Accession GSM5122708 SRX10202445 GSM5122709 SRP308746 SRS8348925 GSM5122709 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122709 GSM5122709 GEO Accession GSM5122709 SRX10202446 GSM5122710 SRP308746 SRS8348926 GSM5122710 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122710 GSM5122710 GEO Accession GSM5122710 SRX10202447 GSM5122711 SRP308746 SRS8348927 GSM5122711 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122711 GSM5122711 GEO Accession GSM5122711 SRX10202448 GSM5122712 SRP308746 SRS8348929 GSM5122712 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122712 GSM5122712 GEO Accession GSM5122712 SRX10202449 GSM5122713 SRP308746 SRS8348928 GSM5122713 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122713 GSM5122713 GEO Accession GSM5122713 SRX10202450 GSM5122714 SRP308746 SRS8348930 GSM5122714 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122714 GSM5122714 GEO Accession GSM5122714 SRX10202451 GSM5122715 SRP308746 SRS8348931 GSM5122715 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122715 GSM5122715 GEO Accession GSM5122715 SRX10202452 GSM5122716 SRP308746 SRS8348932 GSM5122716 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122716 GSM5122716 GEO Accession GSM5122716 SRX10202453 GSM5122717 SRP308746 SRS8348933 GSM5122717 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122717 GSM5122717 GEO Accession GSM5122717 SRX10202454 GSM5122718 SRP308746 SRS8348934 GSM5122718 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122718 GSM5122718 GEO Accession GSM5122718 SRX10202455 GSM5122719 SRP308746 SRS8348935 GSM5122719 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122719 GSM5122719 GEO Accession GSM5122719 SRX10202456 GSM5122720 SRP308746 SRS8348937 GSM5122720 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122720 GSM5122720 GEO Accession GSM5122720 SRX10202457 GSM5122721 SRP308746 SRS8348936 GSM5122721 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122721 GSM5122721 GEO Accession GSM5122721 SRX10202458 GSM5122722 SRP308746 SRS8348938 GSM5122722 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122722 GSM5122722 GEO Accession GSM5122722 SRX10202459 GSM5122723 SRP308746 SRS8348939 GSM5122723 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122723 GSM5122723 GEO Accession GSM5122723 SRX10202460 GSM5122724 SRP308746 SRS8348940 GSM5122724 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122724 GSM5122724 GEO Accession GSM5122724 SRX10202461 GSM5122725 SRP308746 SRS8348943 GSM5122725 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122725 GSM5122725 GEO Accession GSM5122725 SRX10202462 GSM5122726 SRP308746 SRS8348941 GSM5122726 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122726 GSM5122726 GEO Accession GSM5122726 SRX10202463 GSM5122727 SRP308746 SRS8348942 GSM5122727 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122727 GSM5122727 GEO Accession GSM5122727 SRX10202464 GSM5122728 SRP308746 SRS8348944 GSM5122728 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122728 GSM5122728 GEO Accession GSM5122728 SRX10202465 GSM5122729 SRP308746 SRS8348945 GSM5122729 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122729 GSM5122729 GEO Accession GSM5122729 SRX10202466 GSM5122730 SRP308746 SRS8348946 GSM5122730 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122730 GSM5122730 GEO Accession GSM5122730 SRX10202467 GSM5122731 SRP308746 SRS8348948 GSM5122731 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122731 GSM5122731 GEO Accession GSM5122731 SRX10202468 GSM5122732 SRP308746 SRS8348949 GSM5122732 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122732 GSM5122732 GEO Accession GSM5122732 SRX10202469 GSM5122733 SRP308746 SRS8348947 GSM5122733 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122733 GSM5122733 GEO Accession GSM5122733 SRX10202470 GSM5122734 SRP308746 SRS8348950 GSM5122734 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122734 GSM5122734 GEO Accession GSM5122734 SRX10202471 GSM5122735 SRP308746 SRS8348953 GSM5122735 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122735 GSM5122735 GEO Accession GSM5122735 SRX10202472 GSM5122736 SRP308746 SRS8348952 GSM5122736 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122736 GSM5122736 GEO Accession GSM5122736 SRX10202473 GSM5122737 SRP308746 SRS8348951 GSM5122737 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122737 GSM5122737 GEO Accession GSM5122737 SRX10202474 GSM5122738 SRP308746 SRS8348954 GSM5122738 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122738 GSM5122738 GEO Accession GSM5122738 SRX10202475 GSM5122739 SRP308746 SRS8348955 GSM5122739 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122739 GSM5122739 GEO Accession GSM5122739 SRX10202476 GSM5122740 SRP308746 SRS8348956 GSM5122740 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122740 GSM5122740 GEO Accession GSM5122740 SRX10202477 GSM5122741 SRP308746 SRS8348957 GSM5122741 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122741 GSM5122741 GEO Accession GSM5122741 SRX10202478 GSM5122742 SRP308746 SRS8348958 GSM5122742 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122742 GSM5122742 GEO Accession GSM5122742 SRX10202479 GSM5122743 SRP308746 SRS8348959 GSM5122743 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122743 GSM5122743 GEO Accession GSM5122743 SRX10202480 GSM5122744 SRP308746 SRS8348960 GSM5122744 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122744 GSM5122744 GEO Accession GSM5122744 SRX10202481 GSM5122745 SRP308746 SRS8348961 GSM5122745 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122745 GSM5122745 GEO Accession GSM5122745 SRX10202482 GSM5122746 SRP308746 SRS8348963 GSM5122746 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122746 GSM5122746 GEO Accession GSM5122746 SRX10202483 GSM5122747 SRP308746 SRS8348962 GSM5122747 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122747 GSM5122747 GEO Accession GSM5122747 SRX10202484 GSM5122748 SRP308746 SRS8348964 GSM5122748 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122748 GSM5122748 GEO Accession GSM5122748 SRX10202485 GSM5122749 SRP308746 SRS8348965 GSM5122749 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122749 GSM5122749 GEO Accession GSM5122749 SRX10202486 GSM5122750 SRP308746 SRS8348966 GSM5122750 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122750 GSM5122750 GEO Accession GSM5122750 SRX10202487 GSM5122751 SRP308746 SRS8348967 GSM5122751 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122751 GSM5122751 GEO Accession GSM5122751 SRX10202488 GSM5122752 SRP308746 SRS8348968 GSM5122752 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122752 GSM5122752 GEO Accession GSM5122752 SRX10202489 GSM5122753 SRP308746 SRS8348969 GSM5122753 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122753 GSM5122753 GEO Accession GSM5122753 SRX10202490 GSM5122754 SRP308746 SRS8348970 GSM5122754 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122754 GSM5122754 GEO Accession GSM5122754 SRX10202491 GSM5122755 SRP308746 SRS8348972 GSM5122755 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122755 GSM5122755 GEO Accession GSM5122755 SRX10202492 GSM5122756 SRP308746 SRS8348971 GSM5122756 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122756 GSM5122756 GEO Accession GSM5122756 SRX10202493 GSM5122757 SRP308746 SRS8348973 GSM5122757 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122757 GSM5122757 GEO Accession GSM5122757 SRX10202494 GSM5122758 SRP308746 SRS8348974 GSM5122758 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122758 GSM5122758 GEO Accession GSM5122758 SRX10202495 GSM5122759 SRP308746 SRS8348975 GSM5122759 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122759 GSM5122759 GEO Accession GSM5122759 SRX10202496 GSM5122760 SRP308746 SRS8348976 GSM5122760 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122760 GSM5122760 GEO Accession GSM5122760 SRX10202497 GSM5122761 SRP308746 SRS8348977 GSM5122761 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122761 GSM5122761 GEO Accession GSM5122761 SRX10202498 GSM5122762 SRP308746 SRS8348978 GSM5122762 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122762 GSM5122762 GEO Accession GSM5122762 SRX10202499 GSM5122763 SRP308746 SRS8348979 GSM5122763 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122763 GSM5122763 GEO Accession GSM5122763 SRX10202500 GSM5122764 SRP308746 SRS8348980 GSM5122764 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122764 GSM5122764 GEO Accession GSM5122764 SRX10202501 GSM5122765 SRP308746 SRS8348981 GSM5122765 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122765 GSM5122765 GEO Accession GSM5122765 SRX10202502 GSM5122766 SRP308746 SRS8348982 GSM5122766 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122766 GSM5122766 GEO Accession GSM5122766 SRX10202503 GSM5122767 SRP308746 SRS8348983 GSM5122767 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122767 GSM5122767 GEO Accession GSM5122767 SRX10202504 GSM5122768 SRP308746 SRS8348984 GSM5122768 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122768 GSM5122768 GEO Accession GSM5122768 SRX10202505 GSM5122769 SRP308746 SRS8348986 GSM5122769 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122769 GSM5122769 GEO Accession GSM5122769 SRX10202506 GSM5122770 SRP308746 SRS8348985 GSM5122770 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122770 GSM5122770 GEO Accession GSM5122770 SRX10202507 GSM5122771 SRP308746 SRS8348987 GSM5122771 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122771 GSM5122771 GEO Accession GSM5122771 SRX10202508 GSM5122772 SRP308746 SRS8348988 GSM5122772 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122772 GSM5122772 GEO Accession GSM5122772 SRX10202509 GSM5122773 SRP308746 SRS8348989 GSM5122773 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122773 GSM5122773 GEO Accession GSM5122773 SRX10202510 GSM5122774 SRP308746 SRS8348990 GSM5122774 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122774 GSM5122774 GEO Accession GSM5122774 SRX10202511 GSM5122775 SRP308746 SRS8348991 GSM5122775 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122775 GSM5122775 GEO Accession GSM5122775 SRX10202512 GSM5122776 SRP308746 SRS8348992 GSM5122776 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122776 GSM5122776 GEO Accession GSM5122776 SRX10202513 GSM5122777 SRP308746 SRS8348993 GSM5122777 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122777 GSM5122777 GEO Accession GSM5122777 SRX10202514 GSM5122778 SRP308746 SRS8348994 GSM5122778 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122778 GSM5122778 GEO Accession GSM5122778 SRX10202515 GSM5122779 SRP308746 SRS8348995 GSM5122779 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122779 GSM5122779 GEO Accession GSM5122779 SRX10202516 GSM5122780 SRP308746 SRS8348997 GSM5122780 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122780 GSM5122780 GEO Accession GSM5122780 SRX10202517 GSM5122781 SRP308746 SRS8348996 GSM5122781 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122781 GSM5122781 GEO Accession GSM5122781 SRX10202518 GSM5122782 SRP308746 SRS8348998 GSM5122782 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122782 GSM5122782 GEO Accession GSM5122782 SRX10202519 GSM5122783 SRP308746 SRS8349002 GSM5122783 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122783 GSM5122783 GEO Accession GSM5122783 SRX10202520 GSM5122784 SRP308746 SRS8349000 GSM5122784 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122784 GSM5122784 GEO Accession GSM5122784 SRX10202521 GSM5122785 SRP308746 SRS8348999 GSM5122785 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122785 GSM5122785 GEO Accession GSM5122785 SRX10202522 GSM5122786 SRP308746 SRS8349001 GSM5122786 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122786 GSM5122786 GEO Accession GSM5122786 SRX10202523 GSM5122787 SRP308746 SRS8349003 GSM5122787 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122787 GSM5122787 GEO Accession GSM5122787 SRX10202524 GSM5122788 SRP308746 SRS8349005 GSM5122788 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122788 GSM5122788 GEO Accession GSM5122788 SRX10202525 GSM5122789 SRP308746 SRS8349004 GSM5122789 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122789 GSM5122789 GEO Accession GSM5122789 SRX10202526 GSM5122790 SRP308746 SRS8349006 GSM5122790 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122790 GSM5122790 GEO Accession GSM5122790 SRX10202527 GSM5122791 SRP308746 SRS8349007 GSM5122791 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122791 GSM5122791 GEO Accession GSM5122791 SRX10202528 GSM5122792 SRP308746 SRS8349008 GSM5122792 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122792 GSM5122792 GEO Accession GSM5122792 SRX10202529 GSM5122793 SRP308746 SRS8349009 GSM5122793 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122793 GSM5122793 GEO Accession GSM5122793 SRX10202530 GSM5122794 SRP308746 SRS8349010 GSM5122794 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122794 GSM5122794 GEO Accession GSM5122794 SRX10202531 GSM5122795 SRP308746 SRS8349012 GSM5122795 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122795 GSM5122795 GEO Accession GSM5122795 SRX10202532 GSM5122796 SRP308746 SRS8349011 GSM5122796 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122796 GSM5122796 GEO Accession GSM5122796 SRX10202533 GSM5122797 SRP308746 SRS8349013 GSM5122797 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122797 GSM5122797 GEO Accession GSM5122797 SRX10202534 GSM5122798 SRP308746 SRS8349014 GSM5122798 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122798 GSM5122798 GEO Accession GSM5122798 SRX10202535 GSM5122799 SRP308746 SRS8349015 GSM5122799 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122799 GSM5122799 GEO Accession GSM5122799 SRX10202536 GSM5122800 SRP308746 SRS8349016 GSM5122800 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122800 GSM5122800 GEO Accession GSM5122800 SRX10202537 GSM5122801 SRP308746 SRS8349017 GSM5122801 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122801 GSM5122801 GEO Accession GSM5122801 SRX10202538 GSM5122802 SRP308746 SRS8349018 GSM5122802 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122802 GSM5122802 GEO Accession GSM5122802 SRX10202539 GSM5122803 SRP308746 SRS8349020 GSM5122803 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122803 GSM5122803 GEO Accession GSM5122803 SRX10202540 GSM5122804 SRP308746 SRS8349019 GSM5122804 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122804 GSM5122804 GEO Accession GSM5122804 SRX10202541 GSM5122805 SRP308746 SRS8349022 GSM5122805 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122805 GSM5122805 GEO Accession GSM5122805 SRX10202542 GSM5122806 SRP308746 SRS8349021 GSM5122806 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122806 GSM5122806 GEO Accession GSM5122806 SRX10202543 GSM5122807 SRP308746 SRS8349023 GSM5122807 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122807 GSM5122807 GEO Accession GSM5122807 SRX10202544 GSM5122808 SRP308746 SRS8349024 GSM5122808 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122808 GSM5122808 GEO Accession GSM5122808 SRX10202545 GSM5122809 SRP308746 SRS8349025 GSM5122809 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122809 GSM5122809 GEO Accession GSM5122809 SRX10202546 GSM5122810 SRP308746 SRS8349026 GSM5122810 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122810 GSM5122810 GEO Accession GSM5122810 SRX10202547 GSM5122811 SRP308746 SRS8349029 GSM5122811 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122811 GSM5122811 GEO Accession GSM5122811 SRX10202548 GSM5122812 SRP308746 SRS8349027 GSM5122812 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122812 GSM5122812 GEO Accession GSM5122812 SRX10202549 GSM5122813 SRP308746 SRS8349028 GSM5122813 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122813 GSM5122813 GEO Accession GSM5122813 SRX10202550 GSM5122814 SRP308746 SRS8349030 GSM5122814 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122814 GSM5122814 GEO Accession GSM5122814 SRX10202551 GSM5122815 SRP308746 SRS8349031 GSM5122815 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122815 GSM5122815 GEO Accession GSM5122815 SRX10202552 GSM5122816 SRP308746 SRS8349032 GSM5122816 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122816 GSM5122816 GEO Accession GSM5122816 SRX10202553 GSM5122817 SRP308746 SRS8349033 GSM5122817 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122817 GSM5122817 GEO Accession GSM5122817 SRX10202554 GSM5122818 SRP308746 SRS8349035 GSM5122818 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122818 GSM5122818 GEO Accession GSM5122818 SRX10202555 GSM5122819 SRP308746 SRS8349034 GSM5122819 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122819 GSM5122819 GEO Accession GSM5122819 SRX10202556 GSM5122820 SRP308746 SRS8349036 GSM5122820 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122820 GSM5122820 GEO Accession GSM5122820 SRX10202557 GSM5122821 SRP308746 SRS8349039 GSM5122821 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122821 GSM5122821 GEO Accession GSM5122821 SRX10202558 GSM5122822 SRP308746 SRS8349037 GSM5122822 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122822 GSM5122822 GEO Accession GSM5122822 SRX10202559 GSM5122823 SRP308746 SRS8349040 GSM5122823 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122823 GSM5122823 GEO Accession GSM5122823 SRX10202560 GSM5122824 SRP308746 SRS8349041 GSM5122824 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122824 GSM5122824 GEO Accession GSM5122824 SRX10202561 GSM5122825 SRP308746 SRS8349038 GSM5122825 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122825 GSM5122825 GEO Accession GSM5122825 SRX10202562 GSM5122826 SRP308746 SRS8349043 GSM5122826 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122826 GSM5122826 GEO Accession GSM5122826 SRX10202563 GSM5122827 SRP308746 SRS8349044 GSM5122827 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122827 GSM5122827 GEO Accession GSM5122827 SRX10202564 GSM5122828 SRP308746 SRS8349042 GSM5122828 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122828 GSM5122828 GEO Accession GSM5122828 SRX10202565 GSM5122829 SRP308746 SRS8349045 GSM5122829 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122829 GSM5122829 GEO Accession GSM5122829 SRX10202566 GSM5122830 SRP308746 SRS8349048 GSM5122830 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122830 GSM5122830 GEO Accession GSM5122830 SRX10202567 GSM5122831 SRP308746 SRS8349046 GSM5122831 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122831 GSM5122831 GEO Accession GSM5122831 SRX10202568 GSM5122832 SRP308746 SRS8349049 GSM5122832 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122832 GSM5122832 GEO Accession GSM5122832 SRX10202569 GSM5122833 SRP308746 SRS8349047 GSM5122833 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122833 GSM5122833 GEO Accession GSM5122833 SRX10202570 GSM5122834 SRP308746 SRS8349050 GSM5122834 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122834 GSM5122834 GEO Accession GSM5122834 SRX10202571 GSM5122835 SRP308746 SRS8349054 GSM5122835 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122835 GSM5122835 GEO Accession GSM5122835 SRX10202572 GSM5122836 SRP308746 SRS8349052 GSM5122836 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122836 GSM5122836 GEO Accession GSM5122836 SRX10202573 GSM5122837 SRP308746 SRS8349053 GSM5122837 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122837 GSM5122837 GEO Accession GSM5122837 SRX10202574 GSM5122838 SRP308746 SRS8349051 GSM5122838 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122838 GSM5122838 GEO Accession GSM5122838 SRX10202575 GSM5122839 SRP308746 SRS8349055 GSM5122839 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122839 GSM5122839 GEO Accession GSM5122839 SRX10202576 GSM5122840 SRP308746 SRS8349056 GSM5122840 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122840 GSM5122840 GEO Accession GSM5122840 SRX10202577 GSM5122841 SRP308746 SRS8349057 GSM5122841 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122841 GSM5122841 GEO Accession GSM5122841 SRX10202578 GSM5122842 SRP308746 SRS8349058 GSM5122842 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122842 GSM5122842 GEO Accession GSM5122842 SRX10202579 GSM5122843 SRP308746 SRS8349060 GSM5122843 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122843 GSM5122843 GEO Accession GSM5122843 SRX10202580 GSM5122844 SRP308746 SRS8349059 GSM5122844 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122844 GSM5122844 GEO Accession GSM5122844 SRX10202581 GSM5122845 SRP308746 SRS8349061 GSM5122845 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122845 GSM5122845 GEO Accession GSM5122845 SRX10202582 GSM5122846 SRP308746 SRS8349062 GSM5122846 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122846 GSM5122846 GEO Accession GSM5122846 SRX10202583 GSM5122847 SRP308746 SRS8349063 GSM5122847 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122847 GSM5122847 GEO Accession GSM5122847 SRX10202584 GSM5122848 SRP308746 SRS8349065 GSM5122848 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122848 GSM5122848 GEO Accession GSM5122848 SRX10202585 GSM5122849 SRP308746 SRS8349064 GSM5122849 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122849 GSM5122849 GEO Accession GSM5122849 SRX10202586 GSM5122850 SRP308746 SRS8349066 GSM5122850 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122850 GSM5122850 GEO Accession GSM5122850 SRX10202587 GSM5122851 SRP308746 SRS8349067 GSM5122851 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122851 GSM5122851 GEO Accession GSM5122851 SRX10202588 GSM5122852 SRP308746 SRS8349069 GSM5122852 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122852 GSM5122852 GEO Accession GSM5122852 SRX10202589 GSM5122853 SRP308746 SRS8349068 GSM5122853 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122853 GSM5122853 GEO Accession GSM5122853 SRX10202590 GSM5122854 SRP308746 SRS8349070 GSM5122854 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122854 GSM5122854 GEO Accession GSM5122854 SRX10202591 GSM5122855 SRP308746 SRS8349072 GSM5122855 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122855 GSM5122855 GEO Accession GSM5122855 SRX10202592 GSM5122856 SRP308746 SRS8349071 GSM5122856 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122856 GSM5122856 GEO Accession GSM5122856 SRX10202593 GSM5122857 SRP308746 SRS8349074 GSM5122857 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122857 GSM5122857 GEO Accession GSM5122857 SRX10202594 GSM5122858 SRP308746 SRS8349075 GSM5122858 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122858 GSM5122858 GEO Accession GSM5122858 SRX10202595 GSM5122859 SRP308746 SRS8349073 GSM5122859 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122859 GSM5122859 GEO Accession GSM5122859 SRX10202596 GSM5122860 SRP308746 SRS8349076 GSM5122860 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122860 GSM5122860 GEO Accession GSM5122860 SRX10202597 GSM5122861 SRP308746 SRS8349078 GSM5122861 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122861 GSM5122861 GEO Accession GSM5122861 SRX10202598 GSM5122862 SRP308746 SRS8349077 GSM5122862 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122862 GSM5122862 GEO Accession GSM5122862 SRX10202599 GSM5122863 SRP308746 SRS8349079 GSM5122863 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122863 GSM5122863 GEO Accession GSM5122863 SRX10202600 GSM5122864 SRP308746 SRS8349080 GSM5122864 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122864 GSM5122864 GEO Accession GSM5122864 SRX10202601 GSM5122865 SRP308746 SRS8349081 GSM5122865 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122865 GSM5122865 GEO Accession GSM5122865 SRX10202602 GSM5122866 SRP308746 SRS8349082 GSM5122866 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122866 GSM5122866 GEO Accession GSM5122866 SRX10202603 GSM5122867 SRP308746 SRS8349083 GSM5122867 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122867 GSM5122867 GEO Accession GSM5122867 SRX10202604 GSM5122868 SRP308746 SRS8349084 GSM5122868 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122868 GSM5122868 GEO Accession GSM5122868 SRX10202605 GSM5122869 SRP308746 SRS8349085 GSM5122869 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122869 GSM5122869 GEO Accession GSM5122869 SRX10202606 GSM5122870 SRP308746 SRS8349086 GSM5122870 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122870 GSM5122870 GEO Accession GSM5122870 SRX10202607 GSM5122871 SRP308746 SRS8349089 GSM5122871 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122871 GSM5122871 GEO Accession GSM5122871 SRX10202608 GSM5122872 SRP308746 SRS8349088 GSM5122872 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122872 GSM5122872 GEO Accession GSM5122872 SRX10202609 GSM5122873 SRP308746 SRS8349087 GSM5122873 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122873 GSM5122873 GEO Accession GSM5122873 SRX10202610 GSM5122874 SRP308746 SRS8349090 GSM5122874 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122874 GSM5122874 GEO Accession GSM5122874 SRX10202611 GSM5122875 SRP308746 SRS8349091 GSM5122875 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122875 GSM5122875 GEO Accession GSM5122875 SRX10202612 GSM5122876 SRP308746 SRS8349092 GSM5122876 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122876 GSM5122876 GEO Accession GSM5122876 SRX10202613 GSM5122877 SRP308746 SRS8349093 GSM5122877 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122877 GSM5122877 GEO Accession GSM5122877 SRX10202614 GSM5122878 SRP308746 SRS8349094 GSM5122878 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122878 GSM5122878 GEO Accession GSM5122878 SRX10202615 GSM5122879 SRP308746 SRS8349095 GSM5122879 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122879 GSM5122879 GEO Accession GSM5122879 SRX10202616 GSM5122880 SRP308746 SRS8349096 GSM5122880 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122880 GSM5122880 GEO Accession GSM5122880 SRX10202617 GSM5122881 SRP308746 SRS8349098 GSM5122881 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122881 GSM5122881 GEO Accession GSM5122881 SRX10202618 GSM5122882 SRP308746 SRS8349097 GSM5122882 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122882 GSM5122882 GEO Accession GSM5122882 SRX10202619 GSM5122883 SRP308746 SRS8349099 GSM5122883 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122883 GSM5122883 GEO Accession GSM5122883 SRX10202620 GSM5122884 SRP308746 SRS8349100 GSM5122884 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122884 GSM5122884 GEO Accession GSM5122884 SRX10202621 GSM5122885 SRP308746 SRS8349102 GSM5122885 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122885 GSM5122885 GEO Accession GSM5122885 SRX10202622 GSM5122886 SRP308746 SRS8349101 GSM5122886 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122886 GSM5122886 GEO Accession GSM5122886 SRX10202623 GSM5122887 SRP308746 SRS8349105 GSM5122887 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122887 GSM5122887 GEO Accession GSM5122887 SRX10202624 GSM5122888 SRP308746 SRS8349103 GSM5122888 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122888 GSM5122888 GEO Accession GSM5122888 SRX10202625 GSM5122889 SRP308746 SRS8349104 GSM5122889 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122889 GSM5122889 GEO Accession GSM5122889 SRX10202626 GSM5122890 SRP308746 SRS8349106 GSM5122890 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122890 GSM5122890 GEO Accession GSM5122890 SRX10202627 GSM5122891 SRP308746 SRS8349107 GSM5122891 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122891 GSM5122891 GEO Accession GSM5122891 SRX10202628 GSM5122892 SRP308746 SRS8349109 GSM5122892 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122892 GSM5122892 GEO Accession GSM5122892 SRX10202629 GSM5122893 SRP308746 SRS8349108 GSM5122893 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122893 GSM5122893 GEO Accession GSM5122893 SRX10202630 GSM5122894 SRP308746 SRS8349110 GSM5122894 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122894 GSM5122894 GEO Accession GSM5122894 SRX10202631 GSM5122895 SRP308746 SRS8349111 GSM5122895 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122895 GSM5122895 GEO Accession GSM5122895 SRX10202632 GSM5122896 SRP308746 SRS8349112 GSM5122896 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122896 GSM5122896 GEO Accession GSM5122896 SRX10202633 GSM5122897 SRP308746 SRS8349113 GSM5122897 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122897 GSM5122897 GEO Accession GSM5122897 SRX10202634 GSM5122898 SRP308746 SRS8349114 GSM5122898 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122898 GSM5122898 GEO Accession GSM5122898 SRX10202635 GSM5122899 SRP308746 SRS8349116 GSM5122899 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122899 GSM5122899 GEO Accession GSM5122899 SRX10202636 GSM5122900 SRP308746 SRS8349115 GSM5122900 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122900 GSM5122900 GEO Accession GSM5122900 SRX10202637 GSM5122901 SRP308746 SRS8349117 GSM5122901 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122901 GSM5122901 GEO Accession GSM5122901 SRX10202638 GSM5122902 SRP308746 SRS8349118 GSM5122902 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122902 GSM5122902 GEO Accession GSM5122902 SRX10202639 GSM5122903 SRP308746 SRS8349119 GSM5122903 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122903 GSM5122903 GEO Accession GSM5122903 SRX10202640 GSM5122904 SRP308746 SRS8349121 GSM5122904 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122904 GSM5122904 GEO Accession GSM5122904 SRX10202641 GSM5122905 SRP308746 SRS8349120 GSM5122905 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122905 GSM5122905 GEO Accession GSM5122905 SRX10202642 GSM5122906 SRP308746 SRS8349122 GSM5122906 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122906 GSM5122906 GEO Accession GSM5122906 SRX10202643 GSM5122907 SRP308746 SRS8349123 GSM5122907 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122907 GSM5122907 GEO Accession GSM5122907 SRX10202644 GSM5122908 SRP308746 SRS8349126 GSM5122908 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122908 GSM5122908 GEO Accession GSM5122908 SRX10202645 GSM5122909 SRP308746 SRS8349124 GSM5122909 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122909 GSM5122909 GEO Accession GSM5122909 SRX10202646 GSM5122910 SRP308746 SRS8349125 GSM5122910 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122910 GSM5122910 GEO Accession GSM5122910 SRX10202647 GSM5122911 SRP308746 SRS8349127 GSM5122911 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122911 GSM5122911 GEO Accession GSM5122911 SRX10202648 GSM5122912 SRP308746 SRS8349129 GSM5122912 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122912 GSM5122912 GEO Accession GSM5122912 SRX10202649 GSM5122913 SRP308746 SRS8349128 GSM5122913 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122913 GSM5122913 GEO Accession GSM5122913 SRX10202650 GSM5122914 SRP308746 SRS8349130 GSM5122914 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122914 GSM5122914 GEO Accession GSM5122914 SRX10202651 GSM5122915 SRP308746 SRS8349132 GSM5122915 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122915 GSM5122915 GEO Accession GSM5122915 SRX10202652 GSM5122916 SRP308746 SRS8349131 GSM5122916 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122916 GSM5122916 GEO Accession GSM5122916 SRX10202653 GSM5122917 SRP308746 SRS8349133 GSM5122917 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122917 GSM5122917 GEO Accession GSM5122917 SRX10202654 GSM5122918 SRP308746 SRS8349134 GSM5122918 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122918 GSM5122918 GEO Accession GSM5122918 SRX10202655 GSM5122919 SRP308746 SRS8349135 GSM5122919 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122919 GSM5122919 GEO Accession GSM5122919 SRX10202656 GSM5122920 SRP308746 SRS8349136 GSM5122920 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122920 GSM5122920 GEO Accession GSM5122920 SRX10202657 GSM5122921 SRP308746 SRS8349137 GSM5122921 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122921 GSM5122921 GEO Accession GSM5122921 SRX10202658 GSM5122922 SRP308746 SRS8349138 GSM5122922 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122922 GSM5122922 GEO Accession GSM5122922 SRX10202659 GSM5122923 SRP308746 SRS8349139 GSM5122923 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122923 GSM5122923 GEO Accession GSM5122923 SRX10202660 GSM5122924 SRP308746 SRS8349140 GSM5122924 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122924 GSM5122924 GEO Accession GSM5122924 SRX10202661 GSM5122925 SRP308746 SRS8349141 GSM5122925 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122925 GSM5122925 GEO Accession GSM5122925 SRX10202662 GSM5122926 SRP308746 SRS8349143 GSM5122926 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122926 GSM5122926 GEO Accession GSM5122926 SRX10202663 GSM5122927 SRP308746 SRS8349144 GSM5122927 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122927 GSM5122927 GEO Accession GSM5122927 SRX10202664 GSM5122928 SRP308746 SRS8349142 GSM5122928 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122928 GSM5122928 GEO Accession GSM5122928 SRX10202665 GSM5122929 SRP308746 SRS8349145 GSM5122929 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122929 GSM5122929 GEO Accession GSM5122929 SRX10202666 GSM5122930 SRP308746 SRS8349146 GSM5122930 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122930 GSM5122930 GEO Accession GSM5122930 SRX10202667 GSM5122931 SRP308746 SRS8349147 GSM5122931 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122931 GSM5122931 GEO Accession GSM5122931 SRX10202668 GSM5122932 SRP308746 SRS8349148 GSM5122932 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122932 GSM5122932 GEO Accession GSM5122932 SRX10202669 GSM5122933 SRP308746 SRS8349149 GSM5122933 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122933 GSM5122933 GEO Accession GSM5122933 SRX10202670 GSM5122934 SRP308746 SRS8349150 GSM5122934 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122934 GSM5122934 GEO Accession GSM5122934 SRX10202671 GSM5122935 SRP308746 SRS8349151 GSM5122935 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122935 GSM5122935 GEO Accession GSM5122935 SRX10202672 GSM5122936 SRP308746 SRS8349152 GSM5122936 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122936 GSM5122936 GEO Accession GSM5122936 SRX10202673 GSM5122937 SRP308746 SRS8349153 GSM5122937 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122937 GSM5122937 GEO Accession GSM5122937 SRX10202674 GSM5122938 SRP308746 SRS8349154 GSM5122938 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122938 GSM5122938 GEO Accession GSM5122938 SRX10202675 GSM5122939 SRP308746 SRS8349155 GSM5122939 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122939 GSM5122939 GEO Accession GSM5122939 SRX10202676 GSM5122940 SRP308746 SRS8349156 GSM5122940 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122940 GSM5122940 GEO Accession GSM5122940 SRX10202677 GSM5122941 SRP308746 SRS8349159 GSM5122941 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122941 GSM5122941 GEO Accession GSM5122941 SRX10202678 GSM5122942 SRP308746 SRS8349157 GSM5122942 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122942 GSM5122942 GEO Accession GSM5122942 SRX10202679 GSM5122943 SRP308746 SRS8349158 GSM5122943 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122943 GSM5122943 GEO Accession GSM5122943 SRX10202680 GSM5122944 SRP308746 SRS8349160 GSM5122944 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122944 GSM5122944 GEO Accession GSM5122944 SRX10202681 GSM5122945 SRP308746 SRS8349162 GSM5122945 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122945 GSM5122945 GEO Accession GSM5122945 SRX10202682 GSM5122946 SRP308746 SRS8349161 GSM5122946 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122946 GSM5122946 GEO Accession GSM5122946 SRX10202683 GSM5122947 SRP308746 SRS8349163 GSM5122947 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122947 GSM5122947 GEO Accession GSM5122947 SRX10202684 GSM5122948 SRP308746 SRS8349164 GSM5122948 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122948 GSM5122948 GEO Accession GSM5122948 SRX10202685 GSM5122949 SRP308746 SRS8349166 GSM5122949 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122949 GSM5122949 GEO Accession GSM5122949 SRX10202686 GSM5122950 SRP308746 SRS8349165 GSM5122950 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122950 GSM5122950 GEO Accession GSM5122950 SRX10202687 GSM5122951 SRP308746 SRS8349168 GSM5122951 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122951 GSM5122951 GEO Accession GSM5122951 SRX10202688 GSM5122952 SRP308746 SRS8349167 GSM5122952 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122952 GSM5122952 GEO Accession GSM5122952 SRX10202689 GSM5122953 SRP308746 SRS8349170 GSM5122953 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122953 GSM5122953 GEO Accession GSM5122953 SRX10202690 GSM5122954 SRP308746 SRS8349169 GSM5122954 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122954 GSM5122954 GEO Accession GSM5122954 SRX10202691 GSM5122955 SRP308746 SRS8349171 GSM5122955 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122955 GSM5122955 GEO Accession GSM5122955 SRX10202692 GSM5122956 SRP308746 SRS8349172 GSM5122956 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122956 GSM5122956 GEO Accession GSM5122956 SRX10202693 GSM5122957 SRP308746 SRS8349173 GSM5122957 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122957 GSM5122957 GEO Accession GSM5122957 SRX10202694 GSM5122958 SRP308746 SRS8349174 GSM5122958 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122958 GSM5122958 GEO Accession GSM5122958 SRX10202695 GSM5122959 SRP308746 SRS8349175 GSM5122959 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122959 GSM5122959 GEO Accession GSM5122959 SRX10202696 GSM5122960 SRP308746 SRS8349176 GSM5122960 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122960 GSM5122960 GEO Accession GSM5122960 SRX10202697 GSM5122961 SRP308746 SRS8349177 GSM5122961 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122961 GSM5122961 GEO Accession GSM5122961 SRX10202698 GSM5122962 SRP308746 SRS8349178 GSM5122962 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122962 GSM5122962 GEO Accession GSM5122962 SRX10202699 GSM5122963 SRP308746 SRS8349179 GSM5122963 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122963 GSM5122963 GEO Accession GSM5122963 SRX10202700 GSM5122964 SRP308746 SRS8349180 GSM5122964 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122964 GSM5122964 GEO Accession GSM5122964 SRX10202701 GSM5122965 SRP308746 SRS8349182 GSM5122965 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122965 GSM5122965 GEO Accession GSM5122965 SRX10202702 GSM5122966 SRP308746 SRS8349181 GSM5122966 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122966 GSM5122966 GEO Accession GSM5122966 SRX10202703 GSM5122967 SRP308746 SRS8349183 GSM5122967 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122967 GSM5122967 GEO Accession GSM5122967 SRX10202704 GSM5122968 SRP308746 SRS8349184 GSM5122968 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122968 GSM5122968 GEO Accession GSM5122968 SRX10202705 GSM5122969 SRP308746 SRS8349187 GSM5122969 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122969 GSM5122969 GEO Accession GSM5122969 SRX10202706 GSM5122970 SRP308746 SRS8349186 GSM5122970 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122970 GSM5122970 GEO Accession GSM5122970 SRX10202707 GSM5122971 SRP308746 SRS8349185 GSM5122971 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122971 GSM5122971 GEO Accession GSM5122971 SRX10202708 GSM5122972 SRP308746 SRS8349188 GSM5122972 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122972 GSM5122972 GEO Accession GSM5122972 SRX10202709 GSM5122973 SRP308746 SRS8349189 GSM5122973 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122973 GSM5122973 GEO Accession GSM5122973 SRX10202710 GSM5122974 SRP308746 SRS8349190 GSM5122974 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122974 GSM5122974 GEO Accession GSM5122974 SRX10202711 GSM5122975 SRP308746 SRS8349191 GSM5122975 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122975 GSM5122975 GEO Accession GSM5122975 SRX10202712 GSM5122976 SRP308746 SRS8349194 GSM5122976 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122976 GSM5122976 GEO Accession GSM5122976 SRX10202713 GSM5122977 SRP308746 SRS8349192 GSM5122977 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122977 GSM5122977 GEO Accession GSM5122977 SRX10202714 GSM5122978 SRP308746 SRS8349193 GSM5122978 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122978 GSM5122978 GEO Accession GSM5122978 SRX10202715 GSM5122979 SRP308746 SRS8349196 GSM5122979 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122979 GSM5122979 GEO Accession GSM5122979 SRX10202716 GSM5122980 SRP308746 SRS8349195 GSM5122980 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122980 GSM5122980 GEO Accession GSM5122980 SRX10202717 GSM5122981 SRP308746 SRS8349197 GSM5122981 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122981 GSM5122981 GEO Accession GSM5122981 SRX10202718 GSM5122982 SRP308746 SRS8349199 GSM5122982 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122982 GSM5122982 GEO Accession GSM5122982 SRX10202719 GSM5122983 SRP308746 SRS8349198 GSM5122983 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122983 GSM5122983 GEO Accession GSM5122983 SRX10202720 GSM5122984 SRP308746 SRS8349200 GSM5122984 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122984 GSM5122984 GEO Accession GSM5122984 SRX10202721 GSM5122985 SRP308746 SRS8349204 GSM5122985 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122985 GSM5122985 GEO Accession GSM5122985 SRX10202722 GSM5122986 SRP308746 SRS8349201 GSM5122986 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122986 GSM5122986 GEO Accession GSM5122986 SRX10202723 GSM5122987 SRP308746 SRS8349202 GSM5122987 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122987 GSM5122987 GEO Accession GSM5122987 SRX10202724 GSM5122988 SRP308746 SRS8349203 GSM5122988 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122988 GSM5122988 GEO Accession GSM5122988 SRX10202725 GSM5122989 SRP308746 SRS8349207 GSM5122989 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122989 GSM5122989 GEO Accession GSM5122989 SRX10202726 GSM5122990 SRP308746 SRS8349206 GSM5122990 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122990 GSM5122990 GEO Accession GSM5122990 SRX10202727 GSM5122991 SRP308746 SRS8349205 GSM5122991 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122991 GSM5122991 GEO Accession GSM5122991 SRX10202728 GSM5122992 SRP308746 SRS8349208 GSM5122992 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122992 GSM5122992 GEO Accession GSM5122992 SRX10202729 GSM5122993 SRP308746 SRS8349209 GSM5122993 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122993 GSM5122993 GEO Accession GSM5122993 SRX10202730 GSM5122994 SRP308746 SRS8349210 GSM5122994 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122994 GSM5122994 GEO Accession GSM5122994 SRX10202731 GSM5122995 SRP308746 SRS8349211 GSM5122995 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122995 GSM5122995 GEO Accession GSM5122995 SRX10202732 GSM5122996 SRP308746 SRS8349212 GSM5122996 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122996 GSM5122996 GEO Accession GSM5122996 SRX10202733 GSM5122997 SRP308746 SRS8349213 GSM5122997 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122997 GSM5122997 GEO Accession GSM5122997 SRX10202734 GSM5122998 SRP308746 SRS8349214 GSM5122998 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122998 GSM5122998 GEO Accession GSM5122998 SRX10202735 GSM5122999 SRP308746 SRS8349215 GSM5122999 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305122999 GSM5122999 GEO Accession GSM5122999 SRX10202736 GSM5123000 SRP308746 SRS8349216 GSM5123000 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123000 GSM5123000 GEO Accession GSM5123000 SRX10202737 GSM5123001 SRP308746 SRS8349217 GSM5123001 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123001 GSM5123001 GEO Accession GSM5123001 SRX10202738 GSM5123002 SRP308746 SRS8349220 GSM5123002 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123002 GSM5123002 GEO Accession GSM5123002 SRX10202739 GSM5123003 SRP308746 SRS8349218 GSM5123003 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123003 GSM5123003 GEO Accession GSM5123003 SRX10202740 GSM5123004 SRP308746 SRS8349219 GSM5123004 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123004 GSM5123004 GEO Accession GSM5123004 SRX10202741 GSM5123005 SRP308746 SRS8349221 GSM5123005 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123005 GSM5123005 GEO Accession GSM5123005 SRX10202742 GSM5123006 SRP308746 SRS8349222 GSM5123006 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123006 GSM5123006 GEO Accession GSM5123006 SRX10202743 GSM5123007 SRP308746 SRS8349223 GSM5123007 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123007 GSM5123007 GEO Accession GSM5123007 SRX10202744 GSM5123008 SRP308746 SRS8349224 GSM5123008 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123008 GSM5123008 GEO Accession GSM5123008 SRX10202745 GSM5123009 SRP308746 SRS8349226 GSM5123009 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123009 GSM5123009 GEO Accession GSM5123009 SRX10202746 GSM5123010 SRP308746 SRS8349228 GSM5123010 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123010 GSM5123010 GEO Accession GSM5123010 SRX10202747 GSM5123011 SRP308746 SRS8349227 GSM5123011 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123011 GSM5123011 GEO Accession GSM5123011 SRX10202748 GSM5123012 SRP308746 SRS8349225 GSM5123012 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123012 GSM5123012 GEO Accession GSM5123012 SRX10202749 GSM5123013 SRP308746 SRS8349229 GSM5123013 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123013 GSM5123013 GEO Accession GSM5123013 SRX10202750 GSM5123014 SRP308746 SRS8349230 GSM5123014 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123014 GSM5123014 GEO Accession GSM5123014 SRX10202751 GSM5123015 SRP308746 SRS8349231 GSM5123015 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123015 GSM5123015 GEO Accession GSM5123015 SRX10202752 GSM5123016 SRP308746 SRS8349232 GSM5123016 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123016 GSM5123016 GEO Accession GSM5123016 SRX10202753 GSM5123017 SRP308746 SRS8349234 GSM5123017 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123017 GSM5123017 GEO Accession GSM5123017 SRX10202754 GSM5123018 SRP308746 SRS8349233 GSM5123018 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123018 GSM5123018 GEO Accession GSM5123018 SRX10202755 GSM5123019 SRP308746 SRS8349235 GSM5123019 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123019 GSM5123019 GEO Accession GSM5123019 SRX10202756 GSM5123020 SRP308746 SRS8349236 GSM5123020 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123020 GSM5123020 GEO Accession GSM5123020 SRX10202757 GSM5123021 SRP308746 SRS8349237 GSM5123021 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123021 GSM5123021 GEO Accession GSM5123021 SRX10202758 GSM5123022 SRP308746 SRS8349238 GSM5123022 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123022 GSM5123022 GEO Accession GSM5123022 SRX10202759 GSM5123023 SRP308746 SRS8349239 GSM5123023 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123023 GSM5123023 GEO Accession GSM5123023 SRX10202760 GSM5123024 SRP308746 SRS8349240 GSM5123024 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123024 GSM5123024 GEO Accession GSM5123024 SRX10202761 GSM5123025 SRP308746 SRS8349241 GSM5123025 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123025 GSM5123025 GEO Accession GSM5123025 SRX10202762 GSM5123026 SRP308746 SRS8349242 GSM5123026 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123026 GSM5123026 GEO Accession GSM5123026 SRX10202763 GSM5123027 SRP308746 SRS8349243 GSM5123027 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123027 GSM5123027 GEO Accession GSM5123027 SRX10202764 GSM5123028 SRP308746 SRS8349245 GSM5123028 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123028 GSM5123028 GEO Accession GSM5123028 SRX10202765 GSM5123029 SRP308746 SRS8349244 GSM5123029 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123029 GSM5123029 GEO Accession GSM5123029 SRX10202766 GSM5123030 SRP308746 SRS8349246 GSM5123030 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123030 GSM5123030 GEO Accession GSM5123030 SRX10202767 GSM5123031 SRP308746 SRS8349247 GSM5123031 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123031 GSM5123031 GEO Accession GSM5123031 SRX10202768 GSM5123032 SRP308746 SRS8349249 GSM5123032 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123032 GSM5123032 GEO Accession GSM5123032 SRX10202769 GSM5123033 SRP308746 SRS8349248 GSM5123033 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123033 GSM5123033 GEO Accession GSM5123033 SRX10202770 GSM5123034 SRP308746 SRS8349250 GSM5123034 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123034 GSM5123034 GEO Accession GSM5123034 SRX10202771 GSM5123035 SRP308746 SRS8349251 GSM5123035 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123035 GSM5123035 GEO Accession GSM5123035 SRX10202772 GSM5123036 SRP308746 SRS8349252 GSM5123036 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123036 GSM5123036 GEO Accession GSM5123036 SRX10202773 GSM5123037 SRP308746 SRS8349253 GSM5123037 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123037 GSM5123037 GEO Accession GSM5123037 SRX10202774 GSM5123038 SRP308746 SRS8349254 GSM5123038 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123038 GSM5123038 GEO Accession GSM5123038 SRX10202775 GSM5123039 SRP308746 SRS8349255 GSM5123039 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123039 GSM5123039 GEO Accession GSM5123039 SRX10202776 GSM5123040 SRP308746 SRS8349256 GSM5123040 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123040 GSM5123040 GEO Accession GSM5123040 SRX10202777 GSM5123041 SRP308746 SRS8349257 GSM5123041 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123041 GSM5123041 GEO Accession GSM5123041 SRX10202778 GSM5123042 SRP308746 SRS8349258 GSM5123042 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123042 GSM5123042 GEO Accession GSM5123042 SRX10202779 GSM5123043 SRP308746 SRS8349259 GSM5123043 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123043 GSM5123043 GEO Accession GSM5123043 SRX10202780 GSM5123044 SRP308746 SRS8349261 GSM5123044 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123044 GSM5123044 GEO Accession GSM5123044 SRX10202781 GSM5123045 SRP308746 SRS8349260 GSM5123045 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123045 GSM5123045 GEO Accession GSM5123045 SRX10202782 GSM5123046 SRP308746 SRS8349262 GSM5123046 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123046 GSM5123046 GEO Accession GSM5123046 SRX10202783 GSM5123047 SRP308746 SRS8349264 GSM5123047 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123047 GSM5123047 GEO Accession GSM5123047 SRX10202784 GSM5123048 SRP308746 SRS8349265 GSM5123048 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123048 GSM5123048 GEO Accession GSM5123048 SRX10202785 GSM5123049 SRP308746 SRS8349263 GSM5123049 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123049 GSM5123049 GEO Accession GSM5123049 SRX10202786 GSM5123050 SRP308746 SRS8349266 GSM5123050 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123050 GSM5123050 GEO Accession GSM5123050 SRX10202787 GSM5123051 SRP308746 SRS8349267 GSM5123051 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123051 GSM5123051 GEO Accession GSM5123051 SRX10202788 GSM5123052 SRP308746 SRS8349268 GSM5123052 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123052 GSM5123052 GEO Accession GSM5123052 SRX10202789 GSM5123053 SRP308746 SRS8349269 GSM5123053 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123053 GSM5123053 GEO Accession GSM5123053 SRX10202790 GSM5123054 SRP308746 SRS8349270 GSM5123054 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123054 GSM5123054 GEO Accession GSM5123054 SRX10202791 GSM5123055 SRP308746 SRS8349271 GSM5123055 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123055 GSM5123055 GEO Accession GSM5123055 SRX10202792 GSM5123056 SRP308746 SRS8349272 GSM5123056 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123056 GSM5123056 GEO Accession GSM5123056 SRX10202793 GSM5123057 SRP308746 SRS8349275 GSM5123057 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123057 GSM5123057 GEO Accession GSM5123057 SRX10202794 GSM5123058 SRP308746 SRS8349274 GSM5123058 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123058 GSM5123058 GEO Accession GSM5123058 SRX10202795 GSM5123059 SRP308746 SRS8349273 GSM5123059 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123059 GSM5123059 GEO Accession GSM5123059 SRX10202796 GSM5123060 SRP308746 SRS8349276 GSM5123060 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123060 GSM5123060 GEO Accession GSM5123060 SRX10202797 GSM5123061 SRP308746 SRS8349278 GSM5123061 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123061 GSM5123061 GEO Accession GSM5123061 SRX10202798 GSM5123062 SRP308746 SRS8349280 GSM5123062 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123062 GSM5123062 GEO Accession GSM5123062 SRX10202799 GSM5123063 SRP308746 SRS8349277 GSM5123063 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123063 GSM5123063 GEO Accession GSM5123063 SRX10202800 GSM5123064 SRP308746 SRS8349279 GSM5123064 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123064 GSM5123064 GEO Accession GSM5123064 SRX10202801 GSM5123065 SRP308746 SRS8349282 GSM5123065 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123065 GSM5123065 GEO Accession GSM5123065 SRX10202802 GSM5123066 SRP308746 SRS8349283 GSM5123066 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123066 GSM5123066 GEO Accession GSM5123066 SRX10202803 GSM5123067 SRP308746 SRS8349281 GSM5123067 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123067 GSM5123067 GEO Accession GSM5123067 SRX10202804 GSM5123068 SRP308746 SRS8349284 GSM5123068 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123068 GSM5123068 GEO Accession GSM5123068 SRX10202805 GSM5123069 SRP308746 SRS8349285 GSM5123069 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123069 GSM5123069 GEO Accession GSM5123069 SRX10202806 GSM5123070 SRP308746 SRS8349286 GSM5123070 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123070 GSM5123070 GEO Accession GSM5123070 SRX10202807 GSM5123071 SRP308746 SRS8349287 GSM5123071 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123071 GSM5123071 GEO Accession GSM5123071 SRX10202808 GSM5123072 SRP308746 SRS8349289 GSM5123072 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123072 GSM5123072 GEO Accession GSM5123072 SRX10202809 GSM5123073 SRP308746 SRS8349288 GSM5123073 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123073 GSM5123073 GEO Accession GSM5123073 SRX10202810 GSM5123074 SRP308746 SRS8349290 GSM5123074 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123074 GSM5123074 GEO Accession GSM5123074 SRX10202811 GSM5123075 SRP308746 SRS8349292 GSM5123075 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123075 GSM5123075 GEO Accession GSM5123075 SRX10202812 GSM5123076 SRP308746 SRS8349291 GSM5123076 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123076 GSM5123076 GEO Accession GSM5123076 SRX10202813 GSM5123077 SRP308746 SRS8349294 GSM5123077 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123077 GSM5123077 GEO Accession GSM5123077 SRX10202814 GSM5123078 SRP308746 SRS8349293 GSM5123078 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123078 GSM5123078 GEO Accession GSM5123078 SRX10202815 GSM5123079 SRP308746 SRS8349296 GSM5123079 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123079 GSM5123079 GEO Accession GSM5123079 SRX10202816 GSM5123080 SRP308746 SRS8349295 GSM5123080 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123080 GSM5123080 GEO Accession GSM5123080 SRX10202817 GSM5123081 SRP308746 SRS8349297 GSM5123081 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123081 GSM5123081 GEO Accession GSM5123081 SRX10202818 GSM5123082 SRP308746 SRS8349300 GSM5123082 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123082 GSM5123082 GEO Accession GSM5123082 SRX10202819 GSM5123083 SRP308746 SRS8349298 GSM5123083 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123083 GSM5123083 GEO Accession GSM5123083 SRX10202820 GSM5123084 SRP308746 SRS8349299 GSM5123084 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123084 GSM5123084 GEO Accession GSM5123084 SRX10202821 GSM5123085 SRP308746 SRS8349301 GSM5123085 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123085 GSM5123085 GEO Accession GSM5123085 SRX10202822 GSM5123086 SRP308746 SRS8349302 GSM5123086 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123086 GSM5123086 GEO Accession GSM5123086 SRX10202823 GSM5123087 SRP308746 SRS8349303 GSM5123087 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123087 GSM5123087 GEO Accession GSM5123087 SRX10202824 GSM5123088 SRP308746 SRS8349304 GSM5123088 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123088 GSM5123088 GEO Accession GSM5123088 SRX10202825 GSM5123089 SRP308746 SRS8349305 GSM5123089 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123089 GSM5123089 GEO Accession GSM5123089 SRX10202826 GSM5123090 SRP308746 SRS8349310 GSM5123090 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123090 GSM5123090 GEO Accession GSM5123090 SRX10202827 GSM5123091 SRP308746 SRS8349308 GSM5123091 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123091 GSM5123091 GEO Accession GSM5123091 SRX10202828 GSM5123092 SRP308746 SRS8349307 GSM5123092 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123092 GSM5123092 GEO Accession GSM5123092 SRX10202829 GSM5123093 SRP308746 SRS8349306 GSM5123093 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123093 GSM5123093 GEO Accession GSM5123093 SRX10202830 GSM5123094 SRP308746 SRS8349309 GSM5123094 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123094 GSM5123094 GEO Accession GSM5123094 SRX10202831 GSM5123095 SRP308746 SRS8349311 GSM5123095 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123095 GSM5123095 GEO Accession GSM5123095 SRX10202832 GSM5123096 SRP308746 SRS8349312 GSM5123096 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123096 GSM5123096 GEO Accession GSM5123096 SRX10202833 GSM5123097 SRP308746 SRS8349314 GSM5123097 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123097 GSM5123097 GEO Accession GSM5123097 SRX10202834 GSM5123098 SRP308746 SRS8349313 GSM5123098 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123098 GSM5123098 GEO Accession GSM5123098 SRX10202835 GSM5123099 SRP308746 SRS8349317 GSM5123099 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123099 GSM5123099 GEO Accession GSM5123099 SRX10202836 GSM5123100 SRP308746 SRS8349315 GSM5123100 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123100 GSM5123100 GEO Accession GSM5123100 SRX10202837 GSM5123101 SRP308746 SRS8349316 GSM5123101 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123101 GSM5123101 GEO Accession GSM5123101 SRX10202838 GSM5123102 SRP308746 SRS8349318 GSM5123102 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123102 GSM5123102 GEO Accession GSM5123102 SRX10202839 GSM5123103 SRP308746 SRS8349319 GSM5123103 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123103 GSM5123103 GEO Accession GSM5123103 SRX10202840 GSM5123104 SRP308746 SRS8349320 GSM5123104 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123104 GSM5123104 GEO Accession GSM5123104 SRX10202841 GSM5123105 SRP308746 SRS8349322 GSM5123105 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123105 GSM5123105 GEO Accession GSM5123105 SRX10202842 GSM5123106 SRP308746 SRS8349321 GSM5123106 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123106 GSM5123106 GEO Accession GSM5123106 SRX10202843 GSM5123107 SRP308746 SRS8349323 GSM5123107 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123107 GSM5123107 GEO Accession GSM5123107 SRX10202844 GSM5123108 SRP308746 SRS8349324 GSM5123108 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123108 GSM5123108 GEO Accession GSM5123108 SRX10202845 GSM5123109 SRP308746 SRS8349325 GSM5123109 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123109 GSM5123109 GEO Accession GSM5123109 SRX10202846 GSM5123110 SRP308746 SRS8349326 GSM5123110 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123110 GSM5123110 GEO Accession GSM5123110 SRX10202847 GSM5123111 SRP308746 SRS8349328 GSM5123111 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123111 GSM5123111 GEO Accession GSM5123111 SRX10202848 GSM5123112 SRP308746 SRS8349327 GSM5123112 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123112 GSM5123112 GEO Accession GSM5123112 SRX10202849 GSM5123113 SRP308746 SRS8349329 GSM5123113 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123113 GSM5123113 GEO Accession GSM5123113 SRX10202850 GSM5123114 SRP308746 SRS8349330 GSM5123114 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123114 GSM5123114 GEO Accession GSM5123114 SRX10202851 GSM5123115 SRP308746 SRS8349331 GSM5123115 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123115 GSM5123115 GEO Accession GSM5123115 SRX10202852 GSM5123116 SRP308746 SRS8349332 GSM5123116 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123116 GSM5123116 GEO Accession GSM5123116 SRX10202853 GSM5123117 SRP308746 SRS8349333 GSM5123117 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123117 GSM5123117 GEO Accession GSM5123117 SRX10202854 GSM5123118 SRP308746 SRS8349335 GSM5123118 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123118 GSM5123118 GEO Accession GSM5123118 SRX10202855 GSM5123119 SRP308746 SRS8349334 GSM5123119 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123119 GSM5123119 GEO Accession GSM5123119 SRX10202856 GSM5123120 SRP308746 SRS8349336 GSM5123120 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123120 GSM5123120 GEO Accession GSM5123120 SRX10202857 GSM5123121 SRP308746 SRS8349338 GSM5123121 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123121 GSM5123121 GEO Accession GSM5123121 SRX10202858 GSM5123122 SRP308746 SRS8349337 GSM5123122 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123122 GSM5123122 GEO Accession GSM5123122 SRX10202859 GSM5123123 SRP308746 SRS8349339 GSM5123123 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123123 GSM5123123 GEO Accession GSM5123123 SRX10202860 GSM5123124 SRP308746 SRS8349342 GSM5123124 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123124 GSM5123124 GEO Accession GSM5123124 SRX10202861 GSM5123125 SRP308746 SRS8349340 GSM5123125 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123125 GSM5123125 GEO Accession GSM5123125 SRX10202862 GSM5123126 SRP308746 SRS8349341 GSM5123126 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123126 GSM5123126 GEO Accession GSM5123126 SRX10202863 GSM5123127 SRP308746 SRS8349343 GSM5123127 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123127 GSM5123127 GEO Accession GSM5123127 SRX10202864 GSM5123128 SRP308746 SRS8349344 GSM5123128 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123128 GSM5123128 GEO Accession GSM5123128 SRX10202865 GSM5123129 SRP308746 SRS8349345 GSM5123129 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123129 GSM5123129 GEO Accession GSM5123129 SRX10202866 GSM5123130 SRP308746 SRS8349347 GSM5123130 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123130 GSM5123130 GEO Accession GSM5123130 SRX10202867 GSM5123131 SRP308746 SRS8349348 GSM5123131 RNA-Seq TRANSCRIPTOMIC cDNA For mice samples, collected brains were mechanically and enzymatically dissociated using a papain-based brain tumor dissociation kit (Miltenyi Biotec) supplemented with 0.1% typeI collagenase (Thermo Fisher Scientific) / PBS. The dissociated cells were first stained with calcein Blue AM (Thermo Fisher Scientific) and Zombie NIR (BioLegend) for 25 min at 4 ℃, and with anti-mouse CD16/32 (BD Biosciences) for 5min. After washing cells with ice-cold 1% BSA / PBS, the cells were then stained with a combination of following antibodies; PerCP anti-mouse CD45 (clone 30-F11, BD Biosciences), phycoerythrin (PE) anti-mPdgfra (clone APA5, BioLegend), Allophycocyanin (APC) anti-mPodoplanin (clone 8.1.1, BioLegend), and APC anti-CD11b (clone M1/70, BioLegend), for 30 min at 4 ℃. Sorting was performed with Becton Dickinson Influx cytometer (Becton Dickinson). Side scatter (SSC) width versus forward scatter (FSC) area, and Trigger Pulse Width versus FSC criteria were used to discriminate doublets and gate only singleton cells. Viable single cells were identified as calcein blue AM positive and Zombine NIR negative. We sorted viable CD45 negative/ GFP positive or CD45 positive/ CD11b positive single-cells into 96-well plates containing 5uL of TCL buffer (QIAGEN) with 1% beta-mercaptoethanol. Plates were frozen immediately after sorting and stored at -80 ℃ prior to whole transcriptome amplification, library preparation and sequencing. Libraries from isolated single cells from mouse model were generated based on the Smart-seq2 protocol (Picelli 2014) with the following modifications. RNA from single cells was first purified with Agencourt RNAClean XP beads (Beckman Coulter) before oligo-dT primed reverse transcription with Maxima reverse transcriptase and locked TSO oligonucleotide, which was followed by 20 cycle PCR amplification using KAPA HiFi HotStart ReadyMix (KAPA Biosystems) with subsequent Agencourt AMPure XP bead purification as described. Libraries were tagmented using the Nextera XT Library Prep kit (Illumina) with custom barcode adapters (sequences available upon request). Libraries from 768 cells with unique barcodes were combined and sequenced with paired-end, 38-base reads, using a NextSeq 500 sequencer (Illumina). Comma-seperated file containing TPM values (transcript-per-million reads) for 6608 genes across 679 cells. For the cell lines, scRNA-seq libraries were generated using the 10X Genomics Chromium Single Cell 3' Kit v2 and the 10x Chromium Controller (10x Genomics) according to the 10X Single Cell 3' v2 protocol. The cell lines were processed as a pool that consisted of 6 total cell lines, 4 of which were used in this study (MGG23, MGG75, MGH143, MGG18). Two of the used cell lines, MGG23 and MGH143 were profiled again in a different pool. Briefly, we generated a single cell suspension of the pool in 0.04% PBS-BSA (95% viability) and loaded approximately 10,500 single cells to the Chromium Controller with a targeted recovery of 6,000 cells. Single cells, reagents and single gel beads containing barcoded oligonucleotides were encapsulated into nanoliter-sized droplets and subjected to reverse transcription. Droplets were broken and the barcoded cDNAs were purified with DynaBeads and amplified by 12 cycles of PCR (98°C for 45 s; [98°C for 20 s, 67°C for 30 s, 72°C for 1 min] x 12; 72°C for 1 min). The amplified cDNA was fragmented, end-repaired, ligated with index adaptors, and size-selected with clean-ups between each step using the SPRIselect Reagent Kit (Beckman Coulter). Quality control of the resulting barcoded libraries was performed with the Agilent TapeStation and by PCR with primers specific to the P5 and P7 sequence (NEBNext Library Quant Kit for Illumina, New England Biolabs). single cell RNA-seq NextSeq 500 gds 305123131 GSM5123131 GEO Accession GSM5123131