Epigenome Programming by H3.3K27M Mutation Creates a Dependence of Pediatric Glioma on SMARCA4

Abstract

Patients with diffuse midline gliomas that are H3K27 altered (DMG) display a dismal prognosis. However, the molecular mechanisms underlying DMG tumorigenesis remain poorly defined. Here we show that SMARCA4, the catalytic subunit of the mammalian SWI/SNF chromatin remodeling complex, is essential for the proliferation, migration, and invasion of DMG cells and tumor growth in patient-derived DMG xenograft models. SMARCA4 colocalizes with SOX10 at gene regulatory elements to control the expression of genes involved in cell growth and the extracellular matrix (ECM). Moreover, SMARCA4 chromatin binding is reduced upon depletion of SOX10 or H3.3K27M, a mutation occurring in about 60% DMG tumors. Furthermore, the SMARCA4 occupancy at enhancers marked by both SOX10 and H3K27 acetylation is reduced the most upon depleting the H3.3K27M mutation. Taken together, our results support a model in which epigenome reprogramming by H3.3K27M creates a dependence on SMARCA4-mediated chromatin remodeling to drive gene expression and the pathogenesis of H3.3K27M DMG.

Significance: DMG is a deadly pediatric glioma currently without effective treatments. We discovered that the chromatin remodeler SMARCA4 is essential for the proliferation of DMG with H3K27M mutation in vitro and in vivo, identifying a potentially novel therapeutic approach to this disease. See related commentary by Beytagh and Weiss, p. 2730. See related article by Panditharatna et al., p. 2880. This article is highlighted in the In This Issue feature, p. 2711.

Figure 1.. H3K27M DMG cells are vulnerable to SMARCA4 depletion.

(A) Effects of SMARCA4 depletion on the proliferation of DMG cells using a competition assay. Patient-derived H3K27M DMG cells (DIPG17, DIPG4 and DIPG36) were infected with lentivirus at MOI~0.3 for sgRNAs targeting the ROSA-26 (sgNeg), SMARCA4 (sgSMARCA4-1 and sgSMARCA4-2), Ezh2 (sgEZH2-1 and sgEZH2-2), PCNA (sgPCNA). GFP, which is expressed on the same plasmid as sgRNA, positive cells at indicated time point following infection were detected by flow cytometry and reported as percentage of GFP positive cells of three days after infection. (B-D) SMARCA4 depletion in DMG cells results in reduced colony formation. H3K27M DMG cells (DIPG6, DIPG17 and SF8628) and GBM cells with WT H3 (GBM22) and H3.3G34V (KNS42) were infected with lentivirus expressing sgSMARCA4-1, sgSMARCA4-2 and negative control (sgNeg). After selection, cells were collected for Western blot analysis using antibodies against SMARCA4 and Tubulin as controls (B). (C) Representative images of colony formation assay for cells in (B). (D) Relative colony number (n=3 wells per sample) from three independent repeats shown (**p<0.01, ***p<0.001, two-tailed Students’ test). (E-F) Depletion of SMARCA4 results in reduced levels of DNA synthesis and increased apoptosis. After SMARCA4 depletion in two H3K27M DMG cells (DIPG6 and DIPG17) and GBM cells (GBM22 and KNS42), cells were pulsed with 20 μM EdU for 4 hours and EdU positive cells were analyzed by flow cytometry (E). Alternatively, cells were stained with pacific blue- Annexin V and PI and detected by flow cytometry (F). The results represent the average and SD from three independent repeats (*p<0.05, **p<0.01, ***p<0.001, two-tailed Students’ test)

Figure 2.. SMARCA4 is important for DMG tumor growth in patient derived xenografts.

(A-D) Exogenous expression of SMARCA4 from the sgSMARCA4-1 resistant construct restores the proliferation defects caused by depleting endogenous SMARCA4. (A) Western blot analysis of SMARCA4 using antibodies against SMARCA4 or the Flag epitope that is fused to the exogenous SMARCA4. An unspecific band of Flag antibody was marked as *. (B-D) Cell proliferation was monitored by cell viability assays (B) and by colony formation assays (C-D). The average and SD of three independent repeats were shown (**p<0.01, ***p<0.001, two-tailed Students’ test). (E-F) Depletion of SMARCA4 inhibits the tumor growth of DMG xenografts in vivo. DIPG17 cells expressing firefly luciferase were infected with lentivirus expressing sgSMARCA4-1 or sgNeg and were implanted in the pons of female mice. After 11 weeks of implantation, tumor growth of DMG xenografts were monitored every 3 to 4 weeks by bioluminescent imaging (BLI) system. Representative BLI at week 22 was shown in (E) and bioluminescent signals at different time point normalized against week 11 at each individual mouse shown in (F) (sgNeg: n= 6 mice, sgSMARCA4-1: n= 5 mice). (*p<0.05, two-tailed Students’ test). Please note that luminescence measurements were stopped once one tumor-bearing mouse was succumbed to the disease. (G) Survival curves of mice implanted with DIPG17 cells expressing sgSMARCA4-1 or sgNeg. Log-rank analyses were performed to calculate the p value between sgNeg and sgSMARCA4-1 groups (sgNeg= 5, sgSMARCA4-1= 6).

Figure 3.. BRM014 disrupts the viability of DMG cells in vitro and in vivo.

(A-B) Effect of PBRM1 depletion on colony formation of DMG cells (DIPG6, SF8628, and DIPG17) and GBM cells (GBM22 and SF9427). Cells infected with high MOI (~2.1) lentivirus expressed sgPBRM1-P7 and sgPBRM1-P12 targeting PBRM1, and negative control (sgNeg) were collected for Western blot analysis (A) and for colony formation assays (B), with the average and SD from three independent experiments shown (**p<0.01, ***p<0.001, two-tailed Students’ test). (C-D) Effects of BRM014 treatments on the proliferation of DMG cells. (C) H3K27M DMG cells (DIPG6, DIPG17, SF8628, and DIPG36) and H3 WT GBM (GBM22 and GBM43) were treated with indicated concentration of BRM014 for 6 days. Cell viability was measured by CellTiter-Blue assays. (D) DIPG6 and DIPG17 cells were treated with BRM014 at different concentration of BRM014 and cell viability at indicated time was measured. The results were the average and SD from three repeats (**p<0.01, ***p<0.001). (E) Effects of BRM014 treatments on the ability of DMG cells to form colonies. Quantification of colony number in DMG cells (DIPG6, DIPG17, SF8628, DIPG4 and DIPG36) and H3 WT GBM (GBM22 and GBM43) after treated with 0.1 μM and 0.5 μM of BRM014 for 12 days. (n=3, * p<0.05, **p<0.01, ***p<0.001, two-tailed Students’ test). (F) BRM014 treatments inhibit DMG tumor growth in vivo. (F) SF8628 cells were subcutaneously implanted to the flank of mice. Mice were treated with either vehicle (DMSO) or BRM014. The average tumor volume at different time was shown in the left, and with tumor volume of each mouse at Day 47 post-tumor cell injection shown in the right (**p<0.01, two-tailed Students’ test, n=7 mice for Vehicle, n=8 for BRM014). (G) Photographs of nude mice and subcutaneous tumor taken from these mice. Animal survival at the indicated times after inoculation. Log-rank analyses were used to calculate the p value between two treatment groups, n=7 mice for Vehicle, n=8 for BRM014. (H-I) BRM014 treatments reduce tumor growth in mice xenografted with the H3.3K27M mouse DMG line. (H) The average tumor volume in mice xenografted subcutaneously with the mouse H3.3K27M DMG cells treated with either vehicle or BRM014. (I) Tumor weight from each mouse after treatments with tumor images shown at the top panel and tumor weight at lower panel (n=9 each group, * p<0.05, **p<0.01, two-tailed Students’ test).

Figure 4.. SMARCA4 regulates the expression of genes involved in extracellular matrix and cell growth.

(A-C) Effects of SMARCA4 depletion in DIPG6 and DIPG17 cells on gene expression. (A) Volcano plots showing the log₂ fold changes of transcripts in DIPG6 (left) and DIPG17 (right) upon SMARCA4 depletion detected by RNA-seq. Significantly up-and down-regulated genes (FDR<0.05, 1.5-fold) were marked in red and blue, respectively. (B) Venn diagram showing down-regulated genes in both DIPG6 and DIPG17 lines, with gene number in each section shown. (C) Gene ontology analysis of biological processes associated with 198 down-regulated genes. (D) Expression of three genes (SFRP2, HSPG2 and COL2A1) involved in extracellular matrix in two DMG lines (DIPG6 and DIPG17) and two GBM lines (GBM22 and KNS42) by RT-qPCR. The average and SD from three independent repeats shown (two-tailed Students’ test, * p<0.05, ***p<0.001) (E) Boxplot of SFRP2 and LTBP4 expression levels in H3K27M DMG (n=35) and non-brainstem pediatric high-grade glioma (NBS-HGG) with WT H3.3 (n=48) tumor tissues (53). P value was calculated by two-tailed Students’ test. (F-G) Effects of SFRP2 depletion on colony formation of DIPG17 cells. DIPG17 cells were infected with lentivirus expressing two sgRNAs targeting SFRP2 (sgSFRP2-1 and sgSFRP2-2) and negative control (sgNeg). Cells were collected for Western blotting analysis using antibodies against SFRP2 and Tubulin (controls) (F) and for colony formation assays (G), with the results from three independent repeats shown (**p<0.01, two-tailed Students’ test). (H-I) Effects of LTBP4 depletion on DIPG17 cells to form colonies. The experiments were performed as described in F-G. (n=3, ***p<0.001, two-tailed Students’ test). (J-K) Effect of SMARCA4 depletion on the migration of DIPG17 cells. DIPG17 cells infected with sgSMARCA4-1, sgSMARCA4-2 and sgNeg were collected four days after infection and used for the 3D migration assay. Migration distance was measured and normalized to day 0 for each sphere (n=6), with representative images at 0 h and 48 h shown with leading edge outlined in red (J) and the average and SD from three independent experiments shown in (K) (**p<0.01, two-tailed Students’ test). Scale bars, 1000 μm. (L-M) Effect of SMARCA4 depletion on the invasion of DIPG17 cells. DIPG17 cells infected with sgSMARCA4-1, sgSMARCA4-2 and sgNeg were collected at day 4 and used for the 3D invasion assay. Representative images at 0 h and 72 h of the invasion assays were shown with leading edge outlined in Red. Scale bars, 1000 μm. Invasion distance was measured and normalized to day 0 for each sphere (n=6). The average and SD from three independent experiments were shown (**p<0.01, ***p<0.001, two-tailed Students’ test).

Figure 5.. SMARCA4 and SOX10 co-localize at promoters and enhancers in DIPG6 and DIPG17 cells.

(A) Heatmap of SMARCA4 CUT&RUN signals at each of the 8734 peaks found in both DIPG6 and DIPG17. These peaks were separated into three groups based on their co-localization with promoters (high levels of H3K4me3), active enhancers (H3K27ac and low level of H3K4me3) and other sites that do not have the properties of promoters and active enhancers. (B) SMARCA4 occupancy at three genes (SFRP2, HSPG2 and COL2A1) were validated by SMARCA4 CUT&RUN-qPCR in DIPG17 cells with/without SMARCA4 KO (sgSMARCA4-1). An intergenic region without SMARCA4 binding was used as a negative site (negative). SMARCA4 CUT&RUN-qPCR signals were normalized against those of IgG at each locus. The average and SD from three independent experiments were shown. (C) Motif analysis on SMARCA4 CUT&RUN peaks 8734 SMARCA4 peaks shared between DIPG6 and DIPG17, with top sequence motif and adjusted p value shown. (D) SMARCA4 co-immunoprecipitated with SOX10 in DIPG17 cells. (E) Heatmap showing SOX10 CUT&RUN signals at each of the 9106 peaks identified in both DIPG6 and DIPG17, with the number of peaks co-localized with promoters, enhancers and other sites shown. (F-I) SMARCA4 co-localizes with SOX10. The SOX10 CUT&RUN signal density at each of the 17580 SMARCA4 peaks in DIPG6 and 17102 SMARCA4 peaks in DIPG17 was calculated and shown in Heatmap (F, H), with the average density of SOX10 CUT&RUN signals at the SMARCA4 peaks in DIPG6 and DIPG17 shown in (G) and (I), respectively. Results from F-I are the average of two independent repeats. (J) Snapshot of SMARCA4 and SOX10 CUT&RUN signals at three gene loci (SFRP2, HSPG2 and COL2A1) in DIPG17 cells.

Figure 6.. SOX10 depletion affects gene expression, SMARCA4 binding and fitness of H3.3K27M DMG cells.

(A-B) Effect of SOX10 depletion on the expression of three genes (SFRP2, HSPG2 and COL2A1). SOX10 was depleted in DIPG6 and DIPG17 by two sgRNAs (sgSOX10-1 and sgSOX10-5), and was detected by Western blot with Tubulin as controls (A). The effects of SOX10 depletion on the expression of three representative genes involved in extracellular matrix pathway (SFRP2, HSPG2 and COL2A1) were analyzed by RT-qPCR (B). (C) SOX10 occupancy at sites in Figure 5J labelled by black bar was analyzed by SOX10 CUT&RUN-qPCR in negative control (sgNeg) and SOX10 depleted (sgSOX10-1) DIPG17 cells. Negative: an intergenic region without a SOX10 peak. SOX10 CUT&RUN-qPCR signals were normalized by IgG CUT&RUN signals, with the average and SD from three independent experiments shown. (D-E) Effect of SOX10 depletion on the proliferation of two DMG cells. After SOX10 depletion in DIPG6 and DIPG17 cells (A), the proliferation of cells was monitored by the cell viability assays (D) and colony formation assay (E). The average and SD of three independent repeats were shown (* p<0.05, **p<0.01, ***p<0.001). (F-I) Effects of SOX10 depletion on the migration and invasion of DIPG17 cells. DIPG17 cells with SOX10 depletion (sgSOX10-1 and sgSOX10-5) were used for 3D migration (F-G) and 3D invasion assay (H-I), with the average and SD from three independent experiments shown (**p<0.01, ***p<0.001, two-tailed Students’ test). (J-K) Depletion of SOX10 in DIPG17 cells affects SMARCA4 binding to chromatin. DIPG17 cells infected with sgNeg, and sgSOX10-1 were collected for SMARCA4 CUT&RUN. The SMARCA4 CUT&RUN signals in cells treated with sgNeg or sgSOX10-1 at each of the SMARCA4 16937 peaks identified in sgNeg cells (Supplementary Fig. S7D) were calculated and presented as the Heatmap (J), with the average of SMARCA4 CUT&RUN signals at 16937 peaks in these cells shown in (K). (L) Snapshot showing SMARCA4 occupancy at three genes (SFRP2, HSPG2 and COL2A1) in DIPG17 cells with and without SOX10 depletion.

Figure 7.. H3.3K27M mutation confers the dependence of H3.3K27M DMG cells on SMARCA4.

(A-D) H3.3K27M depletion compromised the dependence of DIPG17 on SMARCA4. DIPG17 cells with (sgH3.3K27M) or without H3.3K27M depletion (sgNeg) were transduced with sgNeg or sgSMARCA4-1 for SMARCA4. 4 days after SMARCA4 depletion, cells were collected for Western blot analysis on indicated proteins (A), titer blue cell viability assays (B) and colony formation (C-D). The average and SD from three independent experiments were shown (*p<0.05, ***p<0.001, n.s., not significant, two-tailed Students’ test). (E) Effects of H3.3K27M depletion on gene expression in DIPG17 cells. RNA isolated from cells collected in (A) was used for analysis of gene expression analysis of three genes (SFRP2, HSPG2 and COL2A1) by RT-qPCR (* p<0.05, n.s. not significant, two-tailed Students’ test). (F) Snapshots shown SMARCA4 and SOX10 CUT&RUN signals at the SFRP2 locus in DIPG17 cells with or without H3.3K27M. H3.3K27M CUT&RUN signals in DIPG17 cells were also shown. (G) SMARCA4 chromatin occupancy was reduced upon H3.3K27M depletion. SMARCA4 CUT&RUN experiments were performed in DIPG17 cells with (sgH3.3K27M) or without H3.3K27M depletion (sgNeg), and SMARCA4 peaks were divided into two groups (reduced and no-difference) based on SMARCA4 CUT&RUN signals with a cut-off FC (sgH3.3K27M/sgNeg)<0.83 and FDR<0.05. SMARCA4 CUT&RUN peak density at each SMARCA4 peak in each group was presented as the heatmap (G, left), with the average of SMARCA4 signals of all peaks in each group shown in the right. (H) SMARCA4 occupancy at each SOX10 peaks with reduced density and without changes (no-diff) in DIPG17 cells upon depletion of H3.3K27M. SOX10 CUT&RUN was performed in DIPG17 cells with or without H3.3K27M depletion. SOX10 peaks in DIPG17 cells after depletion of H3.3K27M were divided into two groups (reduced and no-difference) based on the same cut off in (G) and represented in Supplementary Fig. S8P. SMARCA4 peak density at each SOX10 peak was calculated and represented as heat map (H, left), with the average SMARCA4 density at each group of SOX10 peaks shown in the right. (I-J) SMARCA4 CUT&RUN density was reduced the most at regions with reduction of both H3K27ac and SOX10 upon depletion of H3.3K27M. Upon H3.3K27M depletion, loci with reduced H3K27ac and SOX10 alone and with reduced H3K27ac and SOX10 were identified. SMARCA4 CUT&RUN density at each locus of these three groups as well as the average ratio FC (sgH3.3K27M/sgNeg) in each group were calculated and represented by heatmap (I) and Boxplot (J), respectively. Results in G, H, I and J represent the average of two biological replicates, with P value from two-tailed Student’s t test indicated. (K) Most loci with the reduction of both SOX10 and H3K27ac upon depleting H3.3K27M in DIPG17 cells are enhancers. Loci with reduced H3K27ac and SOX10 (n=4282), reduced SOX10 only (n=2450), reduced H3K27ac only (n=12867) and those without changes in either SOX10 (n=10748) or H3K27ac (n=7135) were annotated. (L) A model for the dependence of H3.3K27M DMG cells on SMARCA4 for the regulation the expression of genes involved ECM and growth.