BMP pathway activation coincides with high BMP2/7 levels in ACVR1 WT-H3.3K27M pDMG

A. Principal Component Analysis (PCA) of ACVR1 WT (red) and mutant pDMG (blue) samples based on transcriptomic data from cohort 1.

B. Heatmap representing the transcriptomic expression levels of BMP target genes between ACVR1 WT (red) and mutant pDMG (blue) in cohort 1. H3 mutational status is specified as WT (green), mutated on variant H3.1 (brown) or H3.3 (purple). Normalized and centered gene expression levels are color-coded with a blue (low expression) to red (high expression) gradient. Samples in columns are clustered using Euclidean distance. The BMP targets gene list is presented in Supplementary Table 1.

C. TGF-β/BMP pathway activity inferred from a specific genes-response signature using PROGENy algorithm in ACVR1 WT (red) versus mutant pDMG (blue) (cohort 1). No significant difference was observed between both groups.

D. Comparison of the expression level of BMP ligands, antagonists, receptors, and effectors between ACVR1 WT (red) versus mutant pDMG (blue) in cohort 1. p-values are indicated for each gene.

E. Boxplot of BMP ligands expression (vst-normalized) in cohort 1.

F-G. Pattern of expression of BMP2 and 7 in developing brain. Left panel: heatmaps showing relative BMP2 (F) and BMP7 (G) expression by in situ hybridization (ISH) in murine brain across development obtained from the ALLEN Developing Mouse Brain Atlas. Normalized and scaled gene expression levels are color-coded with a yellow (low expression) to red (high expression) gradient. Developmental stage is mentioned in rows with pre-and post-natal stage color-coded in dark and light blue, respectively. Different brain regions are indicated in columns as following: RSP: rostral secondary prosencephalon; Tel: telencephalic vesicle; PedHy: peduncular caudal hypothalamus; P3: prosomere 3; P2: prosomere 2; P1: prosomere 1; M: midbrain; PPH: prepontine hindbrain; PH: pontine hindbrain; PMH: pontomedullary hindbrain; MH: medullary hindbrain (medulla). Right panel: spatiotemporal gene expression data of BMP2 (F) and BMP7 (G) expression from human developing and adult brain samples obtained from the Human Brain Transcriptome32. The vertical line indicates birth at 266 days. Each curve represents a part of the brain as following: NCX: neocortex (dark blue); HIP: hippocampus (light blue); AMY: amygdala (orange); STR: striatum (black); MD: mediodorsal nucleus of the thalamus (green, in bold); CBC: cerebellar cortex (red).

BMP7 induces a specific transcriptomic and phenotypic switch in a H3.3K27M mutant glioma context

A. BMP7 (left) and BMP2 (right) expressions in H3.3WT (green) versus H3.3K27M (purple) SF188 and Res259 cells. Gene expressions were analyzed by QRT-PCR relative to HPRT1 expression. Means +/- std are represented (n = 3). *: p<0.05, ns: non-significant.

B. TGF-β/BMP pathway activity inferred from a specific genes-response signature using the PROGENy algorithm in Res259-H3.3WT (green) and H3.3K27M (purple) after 0, 3 and 24 hr of BMP7 treatment (n=3). *: p<0.05, ns: non-significant.

C. Venn diagram showing the number of differentially expressed genes (DEG) and the corresponding percentages compared to all DEG in each condition: Res259-H3.3WT versus Res259-H3.3K27M without BMP7 treatment (grey), Res259-H3.3WT versus Res259-H3.3WT treated with BMP7 for 24h (green), Res259-H3.3K27M versus Res259-H3.3K27M treated with BMP7 for 24h (purple).

D. Functional enrichment of DEG specifically between Res259-H3.3K27M versus Res259-H3.3K27M treated with BMP7 for 24h. Dots are colored according to their false discovery rate with a blue (lower significance) to yellow (higher significance) gradient and sized by the count number of genes matching the biological process.

E. Heatmap representing the transcriptomic expression levels of genes associated with cell cycle regulation between Res259-H3.3WT (green) and Res259-H3.3K27M (purple) cells, with (dark blue) or without (light blue) BMP7 treatment. Normalized and centered gene expression levels are color-coded with a blue (low expression) to red (high expression) gradient.

F. Flow cytometry analyses of cell cycle in Res259/SF188-H3.3WT and H3.3K27M upon BMP7 treatment. Left panel: representative density plots with outliers (dots) with 5-ethynyl-2ʹ-deoxyuridine (EdU) staining on the y-axis and with DAPI staining on the x-axis for the indicated conditions on Res259 cell lines. Quantification of cells in G0/G1 phase (blue square, low EdU and low DAPI stainings) appear in the lower left corner for the presented graph. Right panel: quantification of cells in G0/G1 phase for SF188- and Res259-H3.3WT or H3.3K27M without BMP7 treatment (light blue) or after 24 hr treatment (dark blue). Means +/- std are represented (n = 3). *: p<0.05, ns: non-significant.

G. CDKN1A (encoding p21) normalized expression from transcriptomic data of Res259-H3.3WT (green) and Res259-H3.3K27M (purple) after 0, 3 or 24 hr of BMP7 treatment. Means +/- std are represented (n = 3). *: p<0.05, ns: non-significant.

H. Western-blot analysis of RB phosphorylation on S780 (pRB) in Res259-H3.3WT or H3.3K27M upon BMP7 treatment. Total RB and β-actin are used as controls. One representative experiment out of 3 is shown.

I. Functional enrichment of DEG specific for the K27M/BMP7 condition, according to the decision tree algorithm presented in Figure 2-figure supplement J. Dots are colored according to their false discovery rate with a blue (lower significance) to yellow (higher significance) gradient and sized by the count number of genes matching the biological process.

J. ITGA2, ROBO2, MMP28 and COL28A1 normalized expression from transcriptomic data of Res259- H3.3WT (green) and Res259-H3.3K27M (purple) after 0, 3 or 24 hr of BMP7 treatment. Means +/- std are represented (n = 3). *: p<0.05, ns: non-significant.

K. Impact of BMP7 treatment on invasion in Res259-H3.3WT versus H3.3K27M. Left panel: representative pictures of a transwell invasion assay of Res259-H3.3WT or H3.3K27M, with and without BMP7 treatment. Scale bar = 3 mm. Right panel: invasion was quantified as the mean value of five independent experiments and represented as a graph. Means +/- std are represented. *: p<0.05, **: p<0.01.

Combined tumor-autonomous BMP2/BMP7 expression drives a quiescent-invasive tumor cell state in pDMG

A. BMP2 expression after hypoxia or ONC201 treatment. Gene expression was analyzed by QRT-PCR relative to HPRT1 expression. Means +/- std are represented (n = 3). *: p<0.05.

B. Growth monitoring of HSJD-DIPG-012 following recombinant BMP2 treatment. Means +/- std are represented (n = 3). *: p<0.05, **: p<0.01, ****: p<0.0001, ns: non-significant.

C. Impact of BMP2 or LDN treatment on KI67-positive staining in HSJD-DIPG-012. Left panel: representative images of Ki67 immunohistochemistry on HSJD-DIPG-012 spheroids treated or not with either BMP2 or LDN-193189. Scale bar = 50 µm. Right panel: quantification of Ki67-positive and negative cells. p-values were computed using Fisher’s exact test. BMP2: 200 ng/mL, LDN-193189: 1 µM.

D. Impact of BMP2 or LDN treatment on tumor cells invasion. Left panel: representative images of HSJD-DIPG-012 spheroids embedded in Matrigel, after 48 hr of BMP2 or LDN-193189 treatment. Scale bar = 250 µm. Right panel: Invasion was quantified as the mean value of four independent experiments and represented as a graph. *: p<0.05. BMP2: 10 ng/mL, LDN-193189: 1 µM.

E. UMAP (uniform manifold approximation and projection) computed on harmony embeddings of the tumor cells of 10 ACVR1 WT-H3.3K27M pDMGs from the scRNA-seq data published by [Jessa et al., 2022]. PROGENy TGF-β/BMP score is colored from blue (low activity score) to yellow (high activity score). Expression of the BMP receptors BMPR1A, BMPR2 and BMPR1B is colored from grey (low expression values) to purple (high expression values).

F. Violin plots of BMP receptors in one (P-1775_S-1775) of the 10 samples based on the TGF-β/BMP PROGENy score. The TGF-β/BMP-low and high groups are colored respectively in light and dark blue. All p-value are lower than 7.357206e-13 (ACVR2A).

G. Dotplot of the first 10 significantly enriched pathways (FDR <= 0.05) in TGF-β/BMP-high cells for each of the 10 ACVR1 WT-H3.3K27M pDMGs, ranked by number of samples with a significant enrichment, using the GO Biological Process database. Only dots of significant enrichments are shown for each sample. Dot color represents the -log10 (p-value) and ranges from blue (high p-value) to yellow (low p-values). Dot size is proportional to the overlap of DE genes and the genes of a geneset.

H. Violin plots of invasion-related genes in one (P-3407_S-3447) of the 10 samples based on the TGF- β/BMP-high/low score. The TGF-β/BMP-low group and high group are respectively colored in light and dark blue. All p-value are lower than 1.34e-8 (HEY1).

I. “Invasive niche” score from [Ren et al., 2023], and associated expression of the BMP receptors BMPR1A, ACVR1 and BMPR2. Color ranges from blue (low score/expression value) to red (high score/expression value).

J. Scatter plot correlating the PROGENy TGF-β/BMP pathway activity score with the “Invasive niche” score for each Visium spot of pDMG Sample-1. The correlation coefficient was computed using Pearson’s method. ****: p-value < 2.e-16.

K. Scaled expression of invasion-related genes in pDMG Sample-1 for each identified area. Dot size represents the proportion of cells expressing the gene. Color ranges from blue (low expression) to red (high expression).