Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2019 Aug 19;10(1):3731.
doi: 10.1038/s41467-019-11493-2.

Mitogenic and Progenitor Gene Programmes in Single Pilocytic Astrocytoma Cells

Affiliations
Free PMC article

Mitogenic and Progenitor Gene Programmes in Single Pilocytic Astrocytoma Cells

Zachary J Reitman et al. Nat Commun. .
Free PMC article

Abstract

Pilocytic astrocytoma (PA), the most common childhood brain tumor, is a low-grade glioma with a single driver BRAF rearrangement. Here, we perform scRNAseq in six PAs using methods that enabled detection of the rearrangement. When compared to higher-grade gliomas, a strikingly higher proportion of the PA cancer cells exhibit a differentiated, astrocyte-like phenotype. A smaller proportion of cells exhibit a progenitor-like phenotype with evidence of proliferation. These express a mitogen-activated protein kinase (MAPK) programme that was absent from higher-grade gliomas. Immune cells, especially microglia, comprise 40% of all cells in the PAs and account for differences in bulk expression profiles between tumor locations and subtypes. These data indicate that MAPK signaling is restricted to relatively undifferentiated cancer cells in PA, with implications for investigational therapies directed at this pathway.

Conflict of interest statement

P.B. and R.B. received grant funding from Novartis and R.B. has received consulting fees from Novartis. R.B. consults for and owns shares in Ampressa. A.R. is a scientific advisory board member for ThermoFisher Scientific, Syros Pharmaceuticals and Driver Group. A.K.S have received compensation for consulting and SAB membership from Honeycomb Biotechnologies, Cellarity, Cogen Therapeutics, and Dahlia Biosciences. M.W.K. is now an employee of Bristol-Myers Squibb. G.B. is now an employee of Roche. All the remaining authors declare no competing interests.

Figures

Fig. 1
Fig. 1
Process for single cell RNA-seq and cancer cell identification in low grade brain tumors. a Workflow for rapid dissociation, A2B5 immunolabeling and flow sorting to enrich for A2B5+ glial progenitor cells, followed by single cell RNA sequencing. b Schematic for BRAF Spike-in-Seq. A 3’ BRAF oligonucleotide (oligo) is added in 1:1 stochiometry with a general 3’ poly-T oligonucleotide to enhance representation of KIAA1549:BRAF in first-strand synthesis pool. First-strand synthesized nucleic acids are then subjected to whole transcriptome amplification, tagmentation, and next-generation sequencing for RNA-seq analysis. Quantitative PCR on the cDNA library is directed at the KIAA1549-BRAF fusion junction to maximize detection of this transcript
Fig. 2
Fig. 2
Clustering of transcriptomic profiles corresponds to A2B5 and KIAA1549:BRAF status of PA cells. a t-SNE plot showing PA single cells colored by six tumors of origin. b t-SNE plot showing PA single cells colored by A2B5 glial progenitor marker status as determined by immunolabeling. c t-SNE plot showing PA single cells colored by KIAA1549:BRAF status as determined by BRAF Spike-in-Seq for cells undergoing quantitative PCR directed at the KIAA1549:BRAF fusion junction. d t-SNE plot showing PA single cells colored by shared nearest neighbors clustering of transcriptomic profiles revealing tumor clusters (0 and 1), a microglia cluster (2), a T cell cluster (3), and a macrophage cluster (4). e Relative expression of glial markers associated with PA OLIG2, APOD, and PDGFRA. Scale shows log-normalized read counts. f Relative expression of markers of immune cells including CCL3 for microglia, SOD2 for macrophages, and IL32 for T cells. Scale shows log-normalized read counts
Fig. 3
Fig. 3
Single cell RNA-seq analysis of immune cells in PA. a Fraction of microglia, macrophages, and T cells identified among A2B5− cells in six tumors subjected to scRNA-seq. b Heat map showing top differentially expressed genes (rows) for PA immune cell and cancer cell clusters. Cells (n = 931, columns) are arranged by cancer cell and immune cell clusters. c Expression of the differential genes from panel b in bulk expression profiles of 151 pediatric low grade gliomas (PLGGs). Histological subtypes are designated as: DA diffuse astrocytoma, DNT dysembryoplastic neuroepithelial tumor, GG ganglioglioma, OD oligodendroglioma, NOS not otherwise specified. d Estimated proportion of tumor and immune cells in 151 pediatric low grade gliomas, based on relative expression of gene signatures for tumor and immune cell types derived from single cell RNA-seq analysis. Center line shows median, hinges show 1st through 3rd quartiles, and whiskers extend to 1.5 times the interquartile range or the value fathest from hinge, whichever is less. e Enrichment of gene signatures derived from comparison between bulk expression profiles of supratentorial and with infratentorial tumors, among cell types in the single-cell dataset. P-values are for Wilcoxon rank sum test. f Expression of immune checkpoint ligand genes CD86, CD80, CD274 encoding PD-L1, and PDCD1LG2 encoding PD-L2 in tumor and immune compartments of six PA tumors analyzed with scRNA-seq. Center line shows median, hinges show 1st through 3rd quartiles, and whiskers extend to minimum and maximum data points
Fig. 4
Fig. 4
Genes and normal brain gene sets expressed by PA cancer cells. a Volcano plot showing differentially expressed genes between tumor and non-tumor clusters. P-values are Wilcoxon rank sum test P-values. Genes in red are FDR < 0.05 with Bonferroni correction. b Heat map showing expression z-scores for top differentially expressed genes between cancer cells and tumor-associated cells (n = 931) in PA. c Heat map showing mean enrichment scores for normal developing midbrain gene signatures for single cells from PA (BRAF-PA), H3K27M-mutated pediatric high-grade midline gliomas (H3K27M), IDH-mutated astrocytomas (IDH-A), and adult IDH-mutated oligodendrogliomas (IDH-O). OMTN oculomotor and trochlear nucleus, Sert serotonergic, NbM medial neuroblast, NbDA neuroblast dopaminergic, DA0-2 dopaminergic neurons, RN red nucleus, Gaba1-2 GABAergic neurons, mNbL1-2 lateral neuroblasts, NbML1-5 mediolateral neuroblasts, NProg neuronal progenitor, Prog progenitor medial floorplate (FPM), lateral floorplate (FPL), midline (M), basal plate (BP); Rgl1-3 radial glia-like cells, Mgl microglia, Endo endothelial cells, Peric pericytes, Epend ependymal, OPC oligodendrocyte precursor cells. d Violin plot showing ProgFPM lineage scores for cancer cells of each tumor type. e Violin plot showing OPC lineage scores for cancer cells of each tumor type. ****P < 0.0001 (Wilcoxon rank sum test)
Fig. 5
Fig. 5
MAPK signaling and glia-like gene programme in distinct subsets of PA cells. a Cancer cells (n = 531, columns) ranked by principal component 1 derived from PCA of cancer cells. Rows indicate the 15 top and 15 bottom genes associated with principal component 1. b Cancer cells (columns) ranked by principal component 2; rows indicate genes associated with principal component 2. c Heat map showing expression of top 20 genes (rows) from MAPK signaling, astrocyte-like, and oligodendrocyte-like gene programme, across cancer cells (columns) ranked by their expression of each programme. d Expression of selected tumor markers in PA cancer cells from panel c. Red boxes mark populations with >2-fold increased mean scaled expression (P < 0.0001 non-parametric Wilcoxon rank sum test) of the indicated genes compared to all other cells in the dataset. e Relative expression of MAPK signaling and oligodendrocyte-like (OC-like) or astrocyte-like (AC-like) programme in all PA cancer cells. Cells expressing a cell cycling programme are shown in red. Cells expressing a senescence signature are in blue
Fig. 6
Fig. 6
Compartment-specific expression of PA genes. Representative images are shown for F-IHC analysis of BT679 (a, c, e, g) and BT906 (b, d, f, h). ab Pseudo H&E staining. Microcystic and fibrillary components of PA biphasic histology are called out by arrows for both tumors. A vessel is called out as a negative control for tumor cells in BT906. cd GFAP F-IHC, with GFAP positivity indicated in green. e–f Olig2 F-IHC is shown with Olig2 positivity indicated in red. gh Overlay of Olig2 and GFAP F-IHC. Rare Ki67 positive cells are also shown in yellow. A blood vessel is outlined in the bottom left of the BT906 panels (b, d, f, h). i Quantification of Olig2 + and of GFAP + cells for six tumors. n = 14158 cells were analyzed for BT618; n = 22443 for BT646; n = 17234 for BT679; n = 5678 for BT801; n = 12674 for BT827; n = 12696 for BT906. j Representative image of RNA in situ hybridization of PA formalin-fixed paraffin-embedded tissue from BT906 showing expression of a top gene in the MAPK signaling gene programme (JUN) and of a top gene in the astrocyte-like gene programme (APOD). Source data are provided as a Source Data file
Fig. 7
Fig. 7
Inferred trajectories between PA cancer cells. PA cancer cells are plotted based on a lineage trajectory inferred from RNA-seq data. a Cancer cells are colored based on pseudotime inferred from cell trajectory, with pseudotime beginning in the branch on the top right and advancing as cells approach the left and bottom branches. b Cells are colored by tumor of origin. c Cells are colored by MAPK signaling gene programme score. d Cells are colored by AC-like gene programme score
Fig. 8
Fig. 8
Relationship between senescence, cell cycling, MAPK, and glia-like programme in PA cancer cells. a Heat map showing (n = 539) PA cancer cells as columns. Rows show lineage score, MAPK signaling gene programme score, cell cycle score, or senescence score, or BRAF expression. Data are row-scaled and arranged by unsupervised hierarchical clustering. b Plots showing indicated signature scores for PA cancer cells (n = 539). Each panel on the bottom left shows a pair of signature scores and/or BRAF expression values. Signature scores are represented as fold-enrichment scores (i.e., mean increase in expression of genes in the signature divided by mean expression of all expressed genes). BRAF expression is log-normalized read counts. Spearman ρ and Bonferroni-corrected q-values are shown for each pairwise comparison
Fig. 9
Fig. 9
Effect of oncogenic BRAF on single mouse neural stem cell transcriptomes. a t-SNE plot showing mNSCs expressing KIAA1549-BRAF, BRAF-V600E, or vector control constructs. b Heat map showing expression of the top 10 differential genes between mNSCs expressing vector control, KIAA1549-BRAF, or BRAF-V600E constructs. Selected C2 MSigDB gene sets that significantly overlapped with the top 50 differentially expressed genes for each group are shown, along with FDR q-values for significance of overlap using hypergeometric test with Bonferroni correction. c Heat map showing number of genes overlapping between list of top 50 genes differentially expressed between mNSC groups and PA-derived gene programme. d Heat map showing −log10 of P-value for significance of overlap between gene lists using a Fisher’s exact test with a Bonferroni correction. e Violin plot showing AC-like gene programme score for mNSCs expressing vector control, KIAA1549-BRAF, or BRAF-V600E constructs. P-values are for Kruskal–Wallis pairwise tests
Fig. 10
Fig. 10
Gene programme comparisons between low-grade and high-grade pediatric brain tumors. a Venn diagram showing overlap between PA AC-like gene programme and H3K27M AC-like gene programme, along with heat map showing average expression of genes from all AC-like programme. P-value represents significance of overlap based on Fisher’s exact test. Heat map shows average expression of genes from either gene programme in PA and H3K27M cells. Color scale represents average expression z-score. b Single cells from PA tumors, intermediate-grade IDH astrocytomas (IDH-O) and IDH oligodendrogliomas (IDH-A), and from high-grade H3K27M pediatric midline gliomas (H3K27M) are plotted based on expression of the PA AC-like gene programme, and of a H3 AC-like gene programme. Spearman ρ and associated P-values are shown for correlation between the two signatures within cells from each tumor type. c Venn diagram showing lack of overlap between PA MAPK signaling gene programme and H3K27M stem/cellcycle gene programme. Heat map shows average expression of genes from either gene programme in PA and H3K27M cells. Color scale represents average expression z-score. d Single cells from PA and H3K27M-mutated pediatric midline gliomas are plotted based on expression of the PA MAPK signaling gene programme, and of the stem/cell-cycle gene programme derived from H3K27M tumors. e Heat map showing enrichment for human developing midbrain cell type signatures among subpopulations of cells in PA and other glioma types. Color scale shows signature enrichment score. f Model for glioma differentiation hierarchies reflecting differences in compartment, abundance of cycling cells, differentiation state, and population structure of PA compared to IDH-O and IDH-A gliomas, and to H3K27M gliomas

Similar articles

See all similar articles

References

    1. Patel AP, et al. Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma. Science. 2014;344:1396–1401. doi: 10.1126/science.1254257. - DOI - PMC - PubMed
    1. Tirosh I, et al. Single-cell RNA-seq supports a developmental hierarchy in human oligodendroglioma. Nature. 2016;539:309–313. doi: 10.1038/nature20123. - DOI - PMC - PubMed
    1. Venteicher Andrew S., Tirosh Itay, Hebert Christine, Yizhak Keren, Neftel Cyril, Filbin Mariella G., Hovestadt Volker, Escalante Leah E., Shaw McKenzie L., Rodman Christopher, Gillespie Shawn M., Dionne Danielle, Luo Christina C., Ravichandran Hiranmayi, Mylvaganam Ravindra, Mount Christopher, Onozato Maristela L., Nahed Brian V., Wakimoto Hiroaki, Curry William T., Iafrate A. John, Rivera Miguel N., Frosch Matthew P., Golub Todd R., Brastianos Priscilla K., Getz Gad, Patel Anoop P., Monje Michelle, Cahill Daniel P., Rozenblatt-Rosen Orit, Louis David N., Bernstein Bradley E., Regev Aviv, Suvà Mario L. Decoupling genetics, lineages, and microenvironment in IDH-mutant gliomas by single-cell RNA-seq. Science. 2017;355(6332):eaai8478. doi: 10.1126/science.aai8478. - DOI - PMC - PubMed
    1. Filbin MG, et al. Developmental and oncogenic programme in H3K27M gliomas dissected by single-cell RNA-seq. Science. 2018;360:331–335. doi: 10.1126/science.aao4750. - DOI - PMC - PubMed
    1. Stokland T, et al. A multivariate analysis of factors determining tumor progression in childhood low-grade glioma: a population-based cohort study (CCLG CNS9702) Neuro-oncology. 2010;12:1257–1268. - PMC - PubMed

Publication types

MeSH terms

Substances

Feedback