Neoplastic and immune single-cell transcriptomics define subgroup-specific intra-tumoral heterogeneity of childhood medulloblastoma

Abstract

Background: Medulloblastoma (MB) is a heterogeneous disease in which neoplastic cells and associated immune cells contribute to disease progression. We aimed to determine the influence of neoplastic and immune cell diversity on MB biology in patient samples and animal models.

Methods: To better characterize cellular heterogeneity in MB we used single-cell RNA sequencing, immunohistochemistry, and deconvolution of transcriptomic data to profile neoplastic and immune populations in patient samples and animal models across childhood MB subgroups.

Results: Neoplastic cells cluster primarily according to individual sample of origin which is influenced by chromosomal copy number variance. Harmony alignment reveals novel MB subgroup/subtype-associated subpopulations that recapitulate neurodevelopmental processes, including photoreceptor and glutamatergic neuron-like cells in molecular subgroups GP3 and GP4, and a specific nodule-associated neuronally differentiated subpopulation in the sonic hedgehog subgroup. We definitively chart the spectrum of MB immune cell infiltrates, which include subpopulations that recapitulate developmentally related neuron-pruning and antigen-presenting myeloid cells. MB cellular diversity matching human samples is mirrored in subgroup-specific mouse models of MB.

Conclusions: These findings provide a clearer understanding of the diverse neoplastic and immune cell subpopulations that constitute the MB microenvironment.

Keywords: immune; medulloblastoma; neoplastic; scRNA-seq; subpopulation.

Fig. 1

Sample-specific clustering in MB scRNA-seq is influenced by the extent of chromosomal copy number gain and loss. (A) Unaligned UMAP projection of 39.946 single-cell expression data from 28 MB patient samples reveals neoplastic clusters and non-neoplastic clusters. (B) Inference of CNVs (inferCNV) in MB neoplastic and non-neoplastic single cells. (C) Copy number gain or loss event count overlaid on to unaligned UMAP projection of MB cells. Abbreviations: CNV, copy number variants; MB, medulloblastoma; scRNA-seq, single-cell RNA sequencing.

Fig. 2

GP3 MB comprised of progenitor and differentiated neoplastic subpopulations including a photoreceptor-differentiated cluster. (A) Harmony alignment of 12 595 GP3 neoplastic cells colored by identified clusters. (B) Subpopulation proportions in each GP3 sample and corresponding MYC amplification and chromosome 8 gain. (C) GP3-C2 photoreceptor differentiated subpopulation marker IMPG2 IHC score across MB subgroups (GP3, n = 7; GP4, n = 11; SHH, n = 10). Representative IHC staining pattern of IMPG2 (brown) in (D) GP3 (IHC score = 3) and (E) GP4 (IHC score = 2) patient samples (scale bars = 20 µm). Deconvoluted (F) GP3-C2 photoreceptor, (G) GP3-C1 neuronally differentiated, (H) GP3-B2 and (I) GP3-B1 progenitor subpopulation fractions in the MAGIC GP3 cohort (GP3-alpha, n = 67; GP3-beta, n = 37; GP3-gamma, n = 37). Association of (J) chromosome 8 gain (gain, n = 25; loss, n = 60; WT, n = 34), and (K) patient survival with MAGIC GP3-B2 fraction. Abbreviations: IHC, immunohistochemistry; MB, medulloblastoma; SHH, sonic hedgehog; WT, wingless.

Fig. 3

GP4 MB neoplastic cell heterogeneity. (A) Harmony alignment of 14 866 GP4 neoplastic cells colored by identified clusters. (B) Subpopulation proportions in each GP4 sample. (C) Association of subtype with deconvoluted fraction of GP4-B1 with MAGIC GP4 subtypes (GP4-alpha, n = 97; GP4-beta, n = 109; GP4-gamma, n = 11). Representative IHC staining pattern of neuron-differentiated subpopulation GP4-C1 marker GRIA2 (brown) in (D) a GP4 patient sample (IHC score = 2.5) and (E) a GP3 patient sample (IHC score = 0.75; scale bars = 20 µm). (F) GP4-C2 subpopulation marker GRIA2 IHC scores across MB subgroups (GP3, n = 7; GP4, n = 11; SHH, n = 10). (G) Deconvoluted GP4-C2 photoreceptor subpopulation fractions (MAGIC GP4 cohort). (H) Association of patient survival with ratio of progenitor to differentiated deconvoluted fractions (MAGIC GP4 cohort). Abbreviations: IHC, immunohistochemistry; MB, medulloblastoma.

Fig. 4

SHH MB neoplastic subpopulations include a nodule-associated neuronally differentiated subpopulation. (A) Harmony alignment of 6044 SHH neoplastic cells colored by identified clusters. (B) Subpopulation proportions in each SHH sample. (C) Heatmap showing enrichment of top neuron-related GO terms across differentiated (program C) subpopulations. (D and E) Representative IHC staining patterns of SHH-C2 subpopulation marker STMN2 (brown) in SHH patient samples. In addition to STMN2 expression in large classic SHH nodules (black arrows), smaller non-nodular clusters of cells were also expressed STMN2 (red arrows), suggesting areas of developing desmoplasia. (F) Representative absence of Ki-67 IHC staining in a large classic SHH nodule (scale bars = 50 µm). Deconvoluted (G) SHH-C2 and (H) SHH-C1 subpopulation fractions across MAGIC subtypes (SHH-alpha, n = 62; SHH-beta, n = 33; SHH-delta, n = 75; SHH-gamma, n = 44). (I) Association of patient survival with deconvoluted SHH-C1 neuron-differentiated fraction (MAGIC SHH cohort). (J) SHH-B1 fractions in MAGIC subtypes. Abbreviations: GO, gene ontology; IHC, immunohistochemistry; MB, medulloblastoma; SHH, sonic hedgehog.

Fig. 5

The immune landscape of MB. (A) 4669 MB tumor-infiltrating immune cells colored by identified clusters. (B) M2-myeloid subpopulation proportions (scRNA-seq cohort; GP3, n = 7; GP4, n = 11; SHH, n = 9). (C) Deconvoluted M2-myeloid (M2-M) subpopulation fractions across all MB subtypes (MAGIC cohort). (D) Representative IHC staining pattern of M2-myeloid subpopulation marker MRC1 (brown) in an SHH patient sample showing accumulation at nodule linings (scale bar = 100 µm). (E) Deconvoluted nonactivated microglia subpopulation fractions across all MB subtypes (MAGIC cohort). Abbreviations: IHC, immunohistochemistry; MB, medulloblastoma; scRNA-seq, single-cell RNA sequencing; SHH, sonic hedgehog.

Fig. 6

MB subgroup-specific mouse model cellular heterogeneity and fidelity with human MB subpopulations Comparison of mouse models (MP, n = 5008; MG, n = 7658; MS, n = 7143) with human MB subgroup (GP3, n = 12 595; GP4, n = 14 194; SHH, n = 6028) scRNA-seq data by, (A) UMAP projection (Harmony aligned) and (B) hierarchical clustering, reveal that mouse model cells cluster most closely their intended human MB subgroups. (C) 5422 GP3 model MP single cells colored by identified clusters, and labeled according to corresponding subgroup human neoplastic subpopulations, and (D) Jaccard index for the top 200 marker genes between MP and human GP3 neoplastic subpopulations. This was repeated for (E and F) GP3 model MG (n = 8144) and (G and H) SHH model MS (n = 12 421). Abbreviations: MB, medulloblastoma; oligo, oligodendrocyte; scRNA-seq, single-cell RNA sequencing; SHH, sonic hedgehog; vasc., vascular.