Skip to main content
Download PDF

H3F3A K27M mutations drive a repressive transcriptome by modulating chromatin accessibility independent of H3K27me3 in Diffuse Midline Glioma

Epigenetics & Chromatin volume 18, Article number: 23 (2025) Cite this article

Abstract

Background

Heterozygous histone H3.3K27M mutation is a primary oncogenic driver of Diffuse Midline Glioma (DMG). H3.3K27M inhibits the Polycomb Repressive Complex 2 (PRC2) methyltransferase activity, leading to global reduction and redistribution of the repressive H3 lysine 27 tri-methylation (H3K27me3). This epigenomic rewiring is thought to promote gliomagenesis, but the precise role of K27M in gene regulation and tumorigenesis remains incompletely understood.

Results

We established isogenic DMG patient-derived cell lines using CRISPR-Cas9 editing to create H3.3 wild-type (WT), H3.3K27M, and combinations with EZH2 and EZH1 co-deletion, thereby eliminating PRC2 function and H3K27me3. RNA-seq and ATAC-seq analysis revealed that K27M exerts a novel epigenetic effect independent of PRC2 inhibition. While PRC2 loss led to widespread gene induction including HOX gene clusters, and activation of biological pathways, K27M induced a balanced gene deregulation with an overall repressive effect on pathway activity. Genes uniquely affected by K27M, independent of PRC2 loss, showed concordant changes in chromatin accessibility, with upregulated genes becoming more accessible. Importantly, xenografts of H3.3K27M/EZH1/2 WT cells formed tumors, whereas /EZH1/2 knockout cells did not, demonstrating a PRC2-independent role of K27M in tumorigenesis.

Conclusion

Our findings reveal that the H3.3K27M mutation alters chromatin accessibility and uniquely deregulates gene expression independent of H3K27 methylation loss. These PRC2-independent functions of K27M contribute to changes in biological pathway activity and are necessary for tumor development, highlighting novel mechanisms of K27M-driven gliomagenesis.

Graphical Abstract

Key points

• We revealed genes regulated by H3.3K27M mutation and PRC2 in DMG.

• H3.3K27M mutation alters chromosome accessibility independent of H3K27me3.

• PRC2-independent effects of K27M mutation are crucial for tumor development.

Importance of the study

This study demonstrates that H3.3 K27M mutations also drive a repressive transcriptome by modulating chromatin accessibility independently of H3K27 trimethylation in Diffuse Midline Glioma (DMG). By isolating the effects of H3.3 K27me3 loss from those of the K27M mutation, we identified common and unique genes and pathways affected by each. We found that genes uniquely deregulated by K27M showed increased chromatin accessibility and upregulated gene expression, unlike other gene subsets affected by PRC2 knockout. Importantly, we determined the PRC2-independent function of K27M is required for tumorigenesis, as xenografts of H3.3 K27M/PRC2 WT cell lines formed tumors, while H3.3WT/PRC2 WT and K27M/PRC2 knockout cells did not. This research builds upon and significantly advances prior studies, such as those identifying EZH2 as a therapeutic target in H3.3K27M DMGs, by revealing critical new pathways for gliomagenesis. The translational significance lies in identifying novel therapeutic targets against this aggressive pediatric cancer.

Introduction

Diffuse Midline Glioma (DMG) poses a significant challenge in pediatric oncology due to its aggressive nature, anatomical location, challenging resection, and resistance to treatment [1]. The survival statistics are stark, with median survival post-diagnosis between 9 and 15 months and less than a 2% five-year survival rate [1, 2]. The identification of heterozygous H3.3 histone A (known as H3-3 A or H3F3A) gene mutations, resulting in the H3.3K27M mutation in about 80% of DMG, marked a critical development in understanding these tumors’ biology [3,4,5,6,7,8].

The Polycomb Repressive Complex 2 (PRC2) is central to epigenetic gene regulation, primarily through the trimethylation of lysine 27 on histone H3 (H3K27me3) [9,10,11,12], which represses genes critical for development, differentiation, and cell proliferation [13,14,15,16]. This modification is predominantly carried out by the catalytic subunit EZH2, but also by EZH1 which is less efficient than EZH2 [12]. Mutations in key members of the PRC2 complex, including enhancer of zeste homolog 2 (EZH2), embryonic ectoderm development (EED), and SUZ12, have been identified across a spectrum of cancers [17, 18]. Gain-of-function mutations in EZH2, for example, contribute to the progression of cancers such as lymphomas, melanoma, prostate, and breast by promoting the trimethylation of H3K27 and silencing tumor suppressor genes [19,20,21,22]. Conversely, loss-of-function mutations in EED and SUZ12 lead to a decrease in H3K27me3 levels, which derepresses oncogenes and promotes tumorigenesis in conditions like malignant peripheral nerve sheath tumors (MPNSTs) and certain gliomas [18, 23].

The H3.3K27M mutation notably diminishes H3K27me3 levels across the genome, although the exact mechanism remains under investigation [24,25,26,27,28,29,30,31,32,33]. While the molecular mechanism linking K27M to H3K27me3 reduction is unclear, recent studies have raised questions about just how impactful the loss of H3K27me3 is to glioma formation. Research using doxycycline-inducible embryonic stem cells expressing either wild-type or mutant H3.3K27M and H3.3K27L variants demonstrated that these mutations lead to immature gene expression profiles during cell differentiation [33]. Crucially, these changes appeared to be independent of PRC2 function, suggesting that the mutation’s impact on gene regulation and chromatin dynamics involves mechanisms beyond traditional epigenetic silencing pathways. While intriguing, this study was carried out in stem cells only and necessitated follow-up work to establish the presence (and importance) of PRC2-independent, K27M-dependent gene regulation in DMG tumors.

In addition, we recently showed that the K27M mutation disrupts phosphorylation of the adjacent Ser31 residue, leading to increased chromosomal instability and missegregation events [34]. Together, these findings suggest that K27M’s impact on gene regulation and chromatin dynamics involves multiple, PRC2-independent mechanisms with implications for chromosomal stability and tumorigenesis.

To dissect the dynamics of DMG tumorigenesis, we generated isogenic DMG cell lines with and without the H3.3K27M mutation, employing precise genome editing with CRISPR-HDR. Following this, cells underwent a second round of CRISPR-Cas9 gene editing to knock out EZH1 and EZH2, thus eliminating the H3K27 methylation capacity of the PRC2 complex. This unique set of isogenic cells—H3.3WT/PRC2WT, H3.3K27M/PRC2WT, H3.3WT/PRC2KO, and H3.3K27M/PRC2KO—allowed us to differentiate the epigenetic changes induced by H3K27M from those resulting from altered H3K27me3 levels or PRC2-independent mechanisms. RNA sequencing and ATAC-Seq analyses provided a comprehensive view of gene expression and chromatin accessibility changes, respectively. Importantly, our findings reveal that K27M has a unique influence on transcription resulting from changes in chromatin state, that are entirely independent of PRC2. Loss of functional PRC2 leads to a systemic induction of gene expression and upregulation of biological pathways, even expressing developmentally repressive genes including HOX gene clusters; whereas K27M leads to a balanced gene deregulation; altering a similar number of up/down regulated genes; but having an overall repressive effect on the biological pathways. In vivo, analyses through xenografts demonstrated that loss of K27M prevented tumor growth; however, loss of EZH1/2 also prevented tumor growth in DMG and adult glioma cells. Together, our genomic analysis and animal studies allowed us to elucidate the roles of the H3.3K27M mutation and PRC2-mediated methylation in cancer biology and, importantly, suggest that K27M has its own epigenetic function(s) (not mediated through PRC2) that is vital for DMG formation.

Materials and methods

Cell lines and culture conditions

The pediatric glioma cell lines used in this study included SF9427 (RRID: CVCL_E5GC), SF8628 (RRID: CVCL_IT46), SF7761 (RRID: CVCL_IT45), and DIPGXVII/SU-DIPG-XVII (RRID: CVCL_C1MW). SF9427 was obtained from the UCSF Medical Center, while SF8628 and SF7761 were acquired from Millipore Sigma. DIPGXVII was kindly gifted by Paul Knoepfler [35]. SF9427 was derived from a frontal cortex biopsy of a 9-year-old female and was first described by [36]. This cell line is histone mutation wild-type (WT) and TP53-WT. SF8628 was derived in situ from the brainstem pons of a 10-year-old female and was first reported by [36]. This cell line harbors the H3.3K27M mutation and is TP53-mutant. DIPG XVII (SU-DIPG-XVII) was derived from a pontine biopsy of a 9-year-old male and was described by [37]. This cell line also harbors the H3.3K27M mutation. SF7761 was derived from a spinal cord biopsy of a 6-year-old female and was also first reported by [36]. U87 and U118 cells were obtained from the ATCC (American Type Culture Collection). U87-MG was originally derived from a 44-year-old male patient diagnosed with glioblastoma (GBM) and is TP53 wild-type, while U118-MG was established from a 50-year-old female patient diagnosed with glioblastoma and harbors a TP53 mutation. SF8628, SF9427, U87 and U118 were cultured at 37 °C at 5% CO₂ in DMEM-high glucose (Sigma, Cat. No. D6546) supplemented with 10% FBS and 1× Penicillin-Streptomycin. SF7761 and DIPGXVII were grown in Tumor Stem Medium [38], which consisted of DMEM/F12 1:1 (Invitrogen), Neurobasal-A (Invitrogen), 10 mM HEPES (Invitrogen), 1× MEM sodium pyruvate (Invitrogen), 1× MEM nonessential amino acids (Invitrogen), 1% GlutaMax (Invitrogen), human basic fibroblast growth factor (20 ng/mL, Shenandoah), human epidermal growth factor (20 ng/mL, Shenandoah), human platelet-derived growth factor (PDGF)-A and PDGF-B (20 ng/mL each, Shenandoah), heparin (10 ng/mL, StemCell Technologies), and B27 supplement without Vitamin A (Invitrogen). DIPGXVII was grown in DMEM with 10% FBS for 96 h before RNA extraction for RNA-seq experiment to minimize differences in gene expression due to growth media.

EZH1 and EZH2 CRISPR/Cas9 editing

EZH1 and EZH2 CRISPR gRNAs were designed using GUIDES software (http://guides.sanjanalab.org/) [39]. gRNAs were cloned into the BsmBI digested linear Lenti-Cas9-gRNA-GFP vector (Addgene #124770) as described earlier (Giuliano et al. 2019). Plasmids were propagated by transforming into Stbl3 E. coli.

CRISPR gRNA sequences

EZH1 gRNA 1: TTGGTAGTTGTACACTTGTG (D72).

EZH2 gRNA 1: TAGCAAAGATGCCTATCCTG (D53).

EZH2 gRNA 2: CAGGATGAAGCAGACAGAAG (D54).

Lentivirus packaging, transduction, and selection of knockout clones

HEK293T cells were transfected with Lenti-Cas9-gRNA-GFP along with packaging plasmids (pMD.2 and PAX) using the calcium phosphate method [40]. 24 h of post transfection media was changed and virus containing supernatant was collected at 48 to 72 h post-transfection. Lentivirus were filtered through a 0.45 μm syringe and transduced the target cells with 4–10 µg/mL polybrene. Culture media was changed after 24 h of transduction. GFP expressed target cells were sorted and single cell colonies were picked and expanded. The knockout clones were confirmed by western blot and sanger sequencing (cell lines and sequencing data are available on request).

H3F3A CRISPR reversion

We previously created of SF8628 H3.3 WT revertant clones [34] using CRISPR/HDR Cas9 editing to generate SF8628 H3.3 WT revertant cells. A custom gRNA targeted the K27M region of H3F3A. This complex, along with recombinant Cas9, H3F3A WT ssDNA, and a GFP-H3F3A WT HDR repair plasmid, was introduced into cells via electroporation. Colonies were screened for GFP expression using an IncuCyte S3 system and verified by Sanger sequencing and allele-specific PCR with agarose gel analysis. Two of the three clones, CD12 and CH4 (Clones 1 and 2), underwent K27M > WT reversion through precise correction to the wild-type sequence. The third clone, B23 (Clone 3), resulted from the deletion of the mutant allele [34].

Western blotting

Western blotting was performed as described previously [41]. See supplementary materials for list of antibodies.

RNA-Seq

Total RNA was extracted using Qiagen’s RNeasy kit and sent to Active motif for 150 bp paired end sequencing by Illumina NovaSeq platform. The analysis of RNA-seq data was conducted using a detailed in-house bioinformatics pipeline. Briefly, the quality of the raw sequencing data was initially assessed by FastQC (version 0.12.1) [42]. Despite the common practice of bypassing trimming in favor of the STAR aligner’s soft-clipping capabilities, we found that trimming adapters and poor-quality bases with fastp (version 0.23.2) [43] before alignment improved the unique read mapping percentage. Sequencing reads were aligned to the human reference genome (GRCh38.p14) using STAR (version 2.7.11a) [44]. The Rsubread package (version 2.16.0), specifically through the featureCounts function [45], was then employed to assign aligned reads to genomic features, converting BAM format alignment files into raw gene read counts. Differential gene expression analysis was performed utilizing DESeq2 (version 1.40.2) [46]. Differentially expressed genes (DEGs) were defined as genes with an absolute log2 fold change ≥ 1.5 and an adjusted p-value ≤ 0.05 (Benjamini-Hochberg correction). Pairwise comparisons were performed between mutant cells with their isogenic WT counterpart as follows:

SF8628 cells
  1. 1.

    SF8628-H3.3K27M-EZH1/2KO vs. SF8628-H3.3WT-EZH1/2WT.

  2. 2.

    SF8628-H3.3WT-EZH1/2KO vs. SF8628-H3.3WT-EZH1/2WT.

  3. 3.

    SF8628-H3.3K27M-EZH1/2WT vs. SF8628-H3.3WT-EZH1/2WT.

These comparisons were designed to delineate the unique effects of K27M mutation from EZH1/2 knockout by using the same isogenic wild-type (WT) reference.

Other cell lines
  1. 4.

    DIPGXVII-H3.3K27M-EZH1/2WT vs. DIPGXVII-H3.3WT-EZH1/2WT.

  2. 5.

    SF9427-H3.3WT-EZH1/2KO vs. SF9427-H3.3WT-EZH1/2WT.

For DIPGXVII and SF9427, only one genetic perturbation (either K27M or EZH1/2 knockout) was present, unlike SF8628, which had both alterations. Genes with low expression levels (mean counts < 10 across all samples) were filtered out to reduce background noise.

Pathway/network analysis/venn diagram

Enriched biological pathways within the differentially expressed genes were determined using gene ontology (GO) analysis in the clusterProfiler package [47]. Data visualization was enhanced through the use of volcano plots with EnhancedVolcano [48] and Venn diagrams via the VennDiagram package [49]. To further evaluate the impact of H3.3K27M mutation on biological pathways, the clusterProfiler package was utilized for gene set enrichment analysis (GSEA) using all the C2 curated gene sets from the Molecular Signature Database (MSigDB). Gene networks for up and down regulated genes were built using STRINGdb package [50]. Network analyses were conducted with a confidence score threshold set at 400 to balance the inclusivity of protein-protein interactions (PPIs) against the risk of false positives. To assess whether the observed network density was statistically significant, we performed a Monte Carlo simulation, comparing the interaction density of our observed networks to randomly sampled gene sets of the same size from the STRING database [50]. For each dataset (K27M-unique and PRC2-unique up- and down-regulated genes), we conducted 1,000 random iterations, retrieving PPIs for each random set and comparing them to the observed networks. The significance of the network density was evaluated by calculating a p-value, representing the proportion of random networks with interaction counts equal to or greater than the observed interactions. Histograms of the interaction distributions were generated to visualize the enrichment of observed networks relative to randomly sampled gene sets.

ATAC-seq analysis

The ATAC-seq analysis was initiated with paired-end 42 (PE42) sequencing reads, obtained through Illumina sequencing. Sequencing reads were aligned to the human reference genome (GRCh38.p14) using Burrows-Wheeler Aligner (BWA mem) algorithm under its default parameters [51]. For further analysis, mapped reads were filtered through Illumina’s purity filter, retaining only the reads with no more than two mismatches, and mapped uniquely to the genome, while removing the PCR duplicates. To identify peaks indicative of open chromatin regions, MACS3 peak calling algorithm was employed [52]. This approach allowed us to pinpoint genomic regions characterized by increased transposition events, termed “Intervals.” To achieve equitable comparisons across samples, normalization techniques were used to adjust tag numbers to match the tag count of the smallest sample within each isogenic cell group. Furthermore, to ensure comprehensive data coverage, intervals overlapping between samples were consolidated into “Merged Regions,” based on the broadest range of overlap. This strategy ensured data comparability, even for sample-specific intervals. Finally, these genomic intervals and merged peak regions underwent annotation to determine their proximity to gene annotations, incorporating assessments of both average and peak fragment densities.

ATAC-seq and RNA-seq integration analysis

To assess the relationship between chromatin accessibility and gene expression changes, we integrated ATAC-seq and RNA-seq datasets by comparing log2 fold changes (log2FC) across conditions. Scatter plots were generated to visualize the correlation between chromatin accessibility (ATAC-seq) and gene expression (RNA-seq) changes. Pearson correlation coefficients (R² values) were calculated to quantify the degree of association between chromatin accessibility and transcriptional changes.

To further investigate chromatin accessibility patterns in differentially expressed genes, we categorized genes as upregulated, downregulated, or unchanged based on RNA-seq results and compared their ATAC-seq peak distributions using violin plots. Dunn’s test was used to assess statistical significance between gene groups.

Data Availability

All raw and processed ATAC-seq and RNA-seq data generated in this study have been deposited in the NCBI Gene Expression Omnibus (GEO) under the accession numbers GSE293770 and GSE293771 respectively.

Results

Establishment of isogenic H3.3K27M and H3.3 WT DMG cell lines

Using CRISPR/Cas9 editing targeting the H3.3K27M region of the H3F3A gene in a DMG patient-derived cell line (SF8628), we generated SF8628 H3.3 wild-type (WT) revertant cells. Sanger sequencing confirmed heterozygosity in GFP-positive clones. Edited clones were also validated with immunoblotting and immunofluorescence, which confirmed the WT reversion of K27M mutants and restoration of H3K27me3 (Fig. 1). Immunoblotting and antibody-based protocols may not be reliable due to the presence of neighboring post-translational modifications (PTMs) that can make it difficult to distinguish and quantify modification states [53,54,55]. So, we further validated our gene editing by mass spectrometry to measure the relative abundance of the most common histone PTMs in H3.3K27 mutant DMG cell lines and a CRISPR revertant control.

Our analysis identified significant differences in histone PTMs between SF8628 K27M mutant DMG cells and CRISPR revertant controls. Reduced methylation at lysine 27 was observed on both H3.1 and H3.3 histone variants in K27M cells (me1: -28.1%, me2: -66.1%, me3: -79.8%, Supplementary Table 1, Figure S2) compared to H3.3 WT revertant clones. Our comparative data from two additional DMG-derived cell lines (the H3.3 WT SF9427 cells, and the H3.3K27M SF7761 cells) emphasizes the pronounced impact of the K27M mutation on histone methylation. We did not observe any significant changes in H3 K27 acetylation, contrast to a trend of increased acetylation reported by other studies [30, 31]. Interestingly, we found that the H3.3K27M mutant constituted 50% of the H3.3 protein in SF7761 cells, suggesting either upregulation or copy number variations.

Fig. 1
figure 1

Genetic Manipulation of PRC2 and H3.3 K27M. Western blot analysis demonstrating CRISPR-Cas9 HDR-mediated reversion of the H3.3K27M mutation to wild-type and the effect of EZH1/EZH2 knockout on H3K27 methylation. Loss of EZH1 and EZH2 abolishes methylation at H3K27. Blots were probed for H3 K27M, EZH1, EZH2, H3K27me3, H3K27me2, H3K27me1, Tubulin, and Total H3. Total H3 serves as a control for histone H3.1, H3.2, and H3.3 methylation, while Tubulin serves as a loading control. H3.3K27M mutant cell lines are highlighted in red, H3.3 wild-type cell lines in green, and EZH1/2 mutant cell lines in blue

Full size image

Lastly, we conducted a growth analysis of all CRISPR clones using continuous monitoring with the IncuCyte S3 in-incubator cell imaging system. Our findings revealed that H3.3 WT revertant cells exhibited a slight but significant reduction in growth compared to parental cells (Figure S1).

Establishment of isogenic EZH1−/− and EZH2−/− DMG cell lines

Using CRISPR/Cas9 editing, we generated EZH1-/- and EZH2-/- knockout SF8628 cells harboring the H3.3K27M mutation, SF8628 cells with wild-type H3.3 (3 pooled clones), and SF9247 cells. After single-cell selection, gene-edited clones were confirmed through immunoblotting, Sanger sequencing, and immunofluorescence (see Fig. 1 and Figure S3). Our results showed that the co-deletion of EZH1 and EZH2 resulted in the elimination of H3 K27me2 and me3. These findings are consistent with those previously reported in EZH1-/- and EZH2-/- cells [10, 56]. Again, we utilized an IncuCyte S3 to measure cell growth and proliferation of all CRISPR clones. EZH1-/- and EZH2-/- cells exhibited a significant reduction in growth in SF8628 WT and K27M mutant cells but not H3.3 WT SF9247 cells (Figure S1). Based on these results three H3.3 mutant and two EZH1/EZH2 clones were selected for analysis.

Analysis of gene expression changes induced by the H3.3K27M mutation and EZH1/EZH2 knockout

RNA and DNA were extracted from the isogenic H3.3 mutant and revertant, EZH1/2 wild type and knockout cell lines as described and characterized above, as well as from paired isogenic DIPGXVII H3.3K27M mutant and wild type revertant cells (kindly gifted by Paul Knoepfler [35]). These samples were analyzed by RNA sequencing and ATAC sequencing. The PCA plot using RNA-seq data (Fig. 2A) demonstrates distinct gene expression profiles across various cell types and mutations, as evident by the clustering primarily based on cell type, with mutation status contributing to further separation within each cell line. To further examine the influence of mutation status within a single cell line, we performed a PCA analysis using only SF8628 cells, which showed moderate separation of samples based on H3.3 and PRC2 status (Figure S4). This suggests that while mutation status impacts transcriptional variation, cell identity remains a dominant driver of clustering. A series of volcano plots visually summarizes the snapshot of transcriptional changes in H3.3K27M and EZH1/2 knockout (KO) mutants compared to their isogenic wild-type counterparts. These plots reveal distinct patterns of gene expression alterations between these mutations. In the volcano plots for the EZH1/2 KO mutants (Fig. 2B, D, F), there is a notable predominance of genes on the right side of the vertical threshold line, indicating a significant upregulation compared to the wild-type. This pattern suggests an induction of gene expression following the loss of the EZH1/2 components of the PRC2 complex, which typically acts to repress gene transcription. Conversely, the volcano plots for the K27M mutants (Fig. 2C, E) display a more balanced distribution of upregulated and downregulated genes. These observations provide preliminary evidence that while the K27M mutation results in balanced deregulation of gene expression, the knockout of EZH1/2 components distinctly leads to gene induction.

Fig. 2
figure 2

Differential Gene Expression in H3.3K27M Mutation and EZH1/2 Knockout vs. Isogenic Wild Type. (A) PCA plot demonstrates that the samples cluster based on cell type and mutation status. (B–F) Volcano plots illustrating RNA-seq results comparing K27M and EZH1/2 KO mutants to their respective isogenic wild-type controls. Log2 fold change (FC) is plotted on the x-axis to show expression changes, and negative log10 of the adjusted p-value (padj) on the y-axis for statistical significance. Red points signify genes with statistically significant changes (padj < 0.05) and a log2 FC beyond ± 1.5. Vertical dashed lines mark the threshold of log2 FC at ± 1.5; red points to the left indicate downregulated genes, and to the right are upregulated genes in mutants relative to wild-type controls. Gene names are annotated to minimize overlap (max.overlaps set to 50), facilitating readability of significant gene markers

Full size image

H3.3K27M mutations cause an overall repression of biological pathways, while their effect on gene expression is balanced

Our transcriptomic analysis revealed distinctive patterns of gene regulation in response to the K27M mutation and EZH1/2 knockouts. Here, ‘dysregulated genes’ refers to genes that are either up- or down-regulated in response to the K27M mutation or EZH1/2 knockout. In SF8628 cells harboring K27M mutation, the ratio of up- and down-regulated genes is 0.99, indicating that K27M mutation leads to a balanced pattern of gene regulation (Supplementary Table 2). In contrast, EZH1/2 knockout has a derepressive effect, with the number of up-regulated genes being 1.4 and 2.5 times higher than down-regulated genes in K27M-EZH1/2KO and K27WT-EZH1/2KO cells, respectively. These findings are consistent and validated in other cell models: SF9427- EZH1/2KO and DIPGXVII-K27M cells when contrasted with their isogenic wild types (Supplementary Table 2).

To better understand these transcriptional changes, we categorized the dysregulated genes using Venn diagrams (Fig. 3A, B) into seven distinct groups, based on their unique or shared dysregulation patterns across K27M and PRC2 knockout contexts. These categories include: (1) genes unique to K27M (2), genes common between K27WT-EZH1/2KO and K27M-EZH1/2KO (3), genes unique to K27M-EZH1/2KO (4), genes unique to K27WT-EZH1/2KO (5), genes common to K27M and K27WT-EZH1/2KO (6), genes common to all three conditions, and (7) genes common to K27M and K27M-EZH1/2KO.

Fig. 3
figure 3

H3.3K27M Mutations Cause an Overall Repression of Biological Pathways while the Effect on Gene Expression is Balanced. Differential gene expression and pathway responses in K27M and PRC2 mutants. (A) Venn diagram illustrating unique and shared down-regulated and (B) up-regulated genes in SF8628 cells with K27M mutation, and EZH1/2KO (PRC2 loss). DEGs were identified using DESeq2 with the following comparisons: K27M-associated: SF8628-H3.3K27M-EZH1/2WT vs. SF8628-H3.3WT-EZH1/2WT. EZH1/2KO associated: SF8628-H3.3WT-EZH1/2KO vs. SF8628-H3.3WT-EZH1/2WT. K27M + EZH1/2KO effects: SF8628-H3.3K27M-EZH1/2KO vs. SF8628-H3.3WT-EZH1/2WT. These comparisons allowed us to delineate distinct transcriptional programs uniquely regulated by H3.3K27M mutation versus PRC2 loss. (C–I) Gene ontology (GO) pathway enrichment analysis of genes unique to each condition or common across conditions, based on the intersections identified in panels A and B. Blue and red indicate down-regulated and up-regulated pathways, respectively. These analyses reveal that genes uniquely affected by K27M mutation are predominantly associated with pathway repression, whereas genes commonly regulated by PRC2 loss tend to drive pathway upregulation, consistent with the repressive function of PRC2

Full size image

Subsequent Gene Ontology (GO) pathway enrichment analysis of these gene categories revealed that the unique effects of K27M largely contribute to pathway repression (Fig. 3C), contrasting with the pathway activation observed in PRC2 knockout (EZH1/2 KO) conditions (Fig. 3D). This suggests that, while K27M indirectly influences gene expression by inhibiting PRC2 function and causing derepression, it also exerts PRC2-independent effects that counterbalance this derepression, leading to an overall balanced gene expression pattern. The repressive effect on pathways observed with K27M-specific genes highlights a novel, independent role of K27M in transcriptional regulation, beyond merely modulating PRC2 function.

In the remaining categories (Fig. 3E–I), we observed distinct pathway responses that varied with the specific gene set and its relationship to K27M mutation and PRC2 loss (EZH1/2 KO). Genes unique to K27M-PRC2KO predominantly drive downregulated pathways (Fig. 3E), while those unique to K27WT-PRC2KO show a balanced mix of pathway activation and repression (Fig. 3F). Pathways associated with genes common between K27WT-PRC2KO and K27M (Fig. 3G), as well as those common to all three conditions (Fig. 3H), also display a mix of up- and downregulation, indicating complex interactions between K27M mutation and PRC2 loss. Notably, genes common between K27M and K27M-PRC2KO primarily contribute to pathway repression (Fig. 3I), suggesting a consistent repressive influence of K27M even when combined with PRC2 loss.

To validate these findings, we conducted Gene Set Enrichment Analysis (GSEA) on all dysregulated genes in DIPGXVII-K27M cells (Figure S5) and SF9427-EZH1/2KO cells (Figure S6) compared to their isogenic wild types. Since the DIPGXVII and SF9427 cell models each include only K27M mutation or PRC2 loss, respectively, these analyses do not differentiate unique K27M effects from PRC2 loss but do support the broader transcriptional trends observed in SF8628 cells. In DIPGXVII cells, GSEA highlights the repressive impact of K27M across pathways, whereas SF9427-EZH1/2KO cells exhibit widespread pathway activation, consistent with PRC2 knockout effects. Together, these data support the novel finding that K27M independently drives pathway repression, while PRC2 loss uniformly alleviates transcriptional repression.

K27M distinctly alters chromatin landscape independent of methylation loss

To investigate whether the gene expression changes observed in K27M and EZH1/2 knockout cells are driven by alterations in chromosome accessibility, we conducted an analysis of our ATAC-seq datasets and integrated with the RNA-seq data. ATAC-seq was selected for its ability to provide a high-resolution, comprehensive view of open chromatin regions without relying on predefined histone marks or specific antibodies, making it well-suited for unbiased profiling of genome-wide accessibility changes. This approach was particularly advantageous for capturing the widespread and selective chromatin accessibility alterations associated with K27M mutation and PRC2 loss, offering deeper insights into the mutation’s impact on chromatin architecture. Scatter plot analyses of log2 fold changes in RNA-seq and ATAC-seq data revealed a moderate correlation across the genome, suggesting that chromatin accessibility plays a partial role in driving gene expression changes induced by these mutations (Fig. 4A–C). Specifically, R-squared values of 0.33, 0.39, and 0.47 for K27M, K27M-EZH1/2, and WT-EZH1/2 cells, respectively, indicate that approximately 33–47% of the variability in gene expression can be associated with changes in chromatin accessibility compared to wild type.

Focusing on the K27M mutation, we examined chromatin accessibility changes specific to genes uniquely dysregulated by K27M and independent of PRC2 loss, as established in our earlier transcriptional analysis (Figure 3). Violin plots demonstrate a significant difference in ATAC-seq peak values between upregulated, downregulated, and unchanged genes within this K27M-unique subset (Figure 4D). Dunn’s test for multiple comparisons confirmed that upregulated genes associated with K27M mutation exhibit significantly higher chromatin accessibility compared to downregulated genes. Furthermore, genes that remained unchanged in expression displayed an intermediate chromatin accessibility profile, reinforcing the directional relationship between gene expression and chromatin accessibility in the context of K27M mutation. This suggests that the K27M mutation exerts direct effects on chromatin accessibility beyond its role in inhibiting PRC2, highlighting an independent pathway by which K27M affects transcription.

Fig. 4
figure 4

K27M alters chromosome accessibility independent of methylation. Integrative analysis of gene expression and chromatin accessibility across SF8628 mutant cells. (A–C) Scatter plot showing the correlation between log2 fold changes in RNA-seq and ATAC-seq data for SF8628 cells with different mutation types (A) SF8628-H3K27M-PRC2WT (B) SF8628-H3K27M-PRC2KO and (C) SF8628-H3K27WT-PRC2KO; each compared to the wild-type. Each point represents an individual gene. The colored lines indicates the linear regression fit, and the R2 value denotes the proportion of variance in ATAC-seq data explained by RNA-seq data. (D–F) Violin plots showing the distribution of median ATAC-seq peak values for genes uniquely dysregulated in each mutation condition: (D) SF8628-H3K27M-PRC2WT, (E) SF8628-H3K27M-PRC2KO, and (F) SF8628-H3K27WT-PRC2KO cells. Downregulated genes are represented in blue, upregulated genes in red, and unchanged genes in green. The white diamond inside the boxplot highlights the median ATAC-seq peak value for each gene category. Statistical significance is indicated by *** (p < 1e-10) and * (p < 1e-3), based on Dunn’s test for multiple comparisons. (G-H) UCSC browser snapshots of ATAC-seq peak values for selected genes including (G) HOX gene clusters, which are upregulated in PRC2 knock outs; and (H) NFIB, HGF, PEX5L, and ABCA8, which are up-regulated in K27M. In each snapshot, K27M, K27WT, K27M-PRC2KO, and WT-PRC2KO are represented in blue, green, red, and orange color respectively

Full size image

In contrast, the EZH1/2 knockout alone does not demonstrate a similar pattern. Genes uniquely dysregulated by EZH1/2 knockout do not show a significant shift in chromatin accessibility between up- and down-regulated genes (Fig. 4E, F). Moreover, unchanged genes in PRC2KO cells exhibit a chromatin accessibility profile indistinguishable from both upregulated and downregulated genes, further confirming that gene expression changes in PRC2KO cells are not accompanied by chromatin accessibility alterations. This finding implies that PRC2 loss has a widespread impact on gene expression that operates independently of chromatin accessibility changes, distinct from the direct chromatin effects observed with K27M.

Further, we observed a pronounced differential impact of EZH1/2 knockout compared to K27M mutation on the chromatin state of HOX gene clusters. PRC2 disruption led to a substantial increase in open chromatin regions across these loci (Fig. 4G), which was paralleled by a significant upregulation of HOX gene expression (Fig. 2B, D, F). In contrast, the K27M mutation did not exhibit a similar capability to induce chromatin relaxation within these genes, a finding that was mirrored in their relatively stable expression levels as per our RNA-seq data. The H3K27M mutation exerts a dominant-negative effect on PRC2 function, inhibiting its enzymatic activity. However, this doesn’t lead to the complete loss of H3K27me3 across the genome. Rather, it results in a global reduction and redistribution of H3K27me3, which might selectively affect gene expression. And it appears that HOX genes (which are crucial regulators of embryonic development and cell differentiation) remain comparatively transcriptionally silent in the presence of the K27M mutation (Fig. 4G). These genes, typically marked by H3K27me3 for repression during development, retain their silenced state, indicating that the reduction of the repressive marks brought about by the K27M mutation is not sufficient to activate their transcription.

Intriguingly, this trend of chromatin and expression modulation by K27M mutation, as opposed to PRC2 knockout, was not universally applied. For instance, genes such as NFIB, HGF, PEX5L, and ABCA8, which are upregulated and thus demonstrate more open chromatin in the presence of K27M mutation, did not follow the same pattern under EZH1/2 knockout (Fig. 4H). This emphasizes the complexity of K27M’s epigenetic regulation, suggesting a selective rather than a broad, indiscriminate influence on chromatin states.

GO enrichment and network visualization reveal divergent cellular pathways in K27M and EZH1/2 KO

Gene Ontology (GO) enrichment analysis reveals a stark contrast between the biological processes impacted by genes uniquely downregulated and those upregulated in K27M mutants. For downregulated genes, enriched pathways highlight a broad range of developmental processes, including embryonic organ development, neural growth, and muscle tissue development (Fig. 5A). In contrast, upregulated genes in K27M mutants are associated with a narrower scope of pathways focused on the extracellular matrix (Fig. 5B). This specificity, particularly in pathways related to extracellular structure organization, suggests that K27M mutation may selectively deregulate pathways that alter cell-matrix interactions, potentially contributing to the invasive properties of gliomas.

The difference in the number of enriched pathways—three for upregulated genes versus 360 for downregulated genes (top 20 shown)—underscores the selective impact of K27M on gene expression. The prominent upregulation of extracellular matrix-related genes, despite an overall trend of repression in other developmental pathways, hints at a dual role for K27M. This mutation not only represses genes essential for regulated cell growth and differentiation but may also activate specific pathways that remodel the tumor microenvironment, thereby supporting a malignant phenotype.

The GO enrichment analysis and gene network visualization for PRC2 loss (EZH1/2 knockout) reveal a profound shift in cellular dynamics. The GO dot plot for upregulated genes showcases a dominance of developmental processes, with key pathways including pattern specification, organ development, and cell fate commitment among the 854 significantly enriched pathways (Fig. 5D). This suggests a reactivation of developmental gene programs typically silenced by PRC2 in differentiated cells. By comparison, the downregulated pathways, though fewer (31 in total), include key structural and stress response elements, such as extracellular matrix organization, indicating a decrease in genes involved in cellular architecture and environmental stress response (Fig. 5C). Network visualizations illustrate the complexity of these changes. The upregulated genes in PRC2 knockout cells form a dense network with numerous interactions (Fig. 5H), suggesting extensive engagement across developmental pathways post-PRC2 loss. In contrast, the network of downregulated genes appears sparse, with fewer interactions, possibly reflecting a loss of cellular structural integrity and diminished stress response capabilities (Fig. 5G). To determine whether the observed network density is biologically meaningful rather than an artifact of gene set size, we performed a Monte Carlo simulation, comparing the interaction density of observed gene networks to randomly sampled gene sets of the same size. The results indicate that, in all cases (K27M-unique and PRC2-unique upregulated and downregulated genes), the observed networks were significantly denser than expected by chance (Supplementary Figure S7). These findings confirm that the observed interactions are unlikely to be random and further support the biological significance of PRC2-dependent transcriptional reprogramming.

Fig. 5
figure 5

GO Enrichment and Network Visualization Reveal Divergent Cellular Pathways in K27M and EZH1/2 KO Gene Ontology (GO) enrichment analysis and network visualization of K27M-unique and EZH1/2 knockout-unique genes in SF8628 cells. (A–D) In the GO enrichment dot plots, each dot represents an enriched GO term, with dot size reflecting the number of genes associated with that term and color indicating statistical significance (p-adjust value). The x-axis shows the GeneRatio, or proportion of genes associated with each GO term relative to the total number of genes analyzed. Only the top 20 significantly enriched biological processes are displayed. (A) Downregulated genes (886 genes) uniquely regulated by K27M mutation. (B) Upregulated genes (813 genes) uniquely regulated by K27M mutation. (C) Downregulated genes (673 genes) uniquely regulated by EZH1/2 knockout. (D) Upregulated genes (1997 genes) uniquely regulated by EZH1/2 knockout. (E–H) Network visualization of uniquely regulated genes in K27M and EZH1/2 knockout SF8628 cells. Nodes represent proteins encoded by differentially expressed genes, while edges indicate protein-protein interactions with a confidence score threshold of 400. Networks were generated using the STRING database, with statistical significance annotated by p-value. (E) Downregulated genes in K27M cells highlight inhibited pathways associated with this mutation. (F) Upregulated genes in K27M cells reveal enhanced protein interaction networks, suggesting active pathways specific to the K27M mutation. (G) Downregulated genes in EZH1/2 knockout cells indicate suppressed pathways due to the loss of EZH1/2 function. (H) Upregulated genes in EZH1/2 knockout cells display a network illustrating the molecular effects of PRC2 disruption on cellular signaling

Full size image

Overall, the loss of PRC2 leads to the reactivation of developmentally programmed genes while simultaneously downregulating genes responsible for structural maintenance and environmental resilience. These findings illustrate distinct, context-dependent roles of K27M mutation and PRC2 loss in altering cellular pathways and gene networks, with K27M selectively activating extracellular matrix pathways and PRC2 loss broadly reawakening developmental processes.

K27M-driven regulatory effects on tumor suppressor genes & oncogenes

Our analysis revealed that CDKN2A was significantly upregulated in K27M mutants compared to their isogenic wild-type controls. Specifically, the SF8628 K27M, SF 8628 K27M EZH1/2KO, and DIPG XVII K27M cell lines showed log2 fold changes of 12.9, 13.4, and 10.9, respectively (Fig. 6). In contrast, the EZH1/2 knockout cells without K27M mutation did not exhibit notable changes in CDKN2A expression, suggesting a K27M-specific regulatory effect independent of PRC2-mediated methylation. ATAC-seq analysis further supports this regulatory effect, as K27M mutant cells exhibited increased chromatin accessibility at the CDKN2A locus, whereas no such accessibility changes were observed in EZH1/2 knockout cells lacking K27M (Fig. 7A). These findings suggest that K27M may exert its transcriptional activation of CDKN2A through chromatin remodeling, rather than through global loss of H3K27me3 alone (Fig. 7A).

Fig. 6
figure 6

K27M-Driven Regulatory Effects on Tumor Suppressor Genes & Oncogenes. (A) Heatmap displaying differential expression of known oncogenes and tumor suppressor genes from the Cancer Gene Census Tier 1 (as retrieved from Cosmic) in K27M and PRC2 mutant cell lines relative to their isogenic wild-type counterparts. The heatmap illustrates log2 fold change (log2FC) values, with red indicating upregulation and blue indicating downregulation in the mutants. Genes included in this figure were stringently selected for a log2FC greater than 1.5. Cells marked with zero indicate absence of data for a given gene in one of the mutation types, replacing non-available (NA) entries for clearer visual representation. This heatmap highlights the oncogenes and tumor suppressor genes most significantly influenced by the mutations

Full size image
Fig. 7
figure 7

K27M-Driven Chromatin Remodeling and Transcriptional Regulation of CDKN2A and HGF. (A) Bar plot showing CDKN2A expression (RNA-seq) and chromatin accessibility (ATAC-seq log2FC) in K27M and PRC2 mutant cells relative to their wild-type counterparts. Increased chromatin accessibility in K27M mutants suggests that K27M-specific transcriptional regulation of CDKN2A is partially mediated through chromatin remodeling. EZH1/2 knockout alone does not alter chromatin accessibility or expression of CDKN2A, indicating that PRC2 loss alone is not sufficient for CDKN2A activation. (B) Bar plot illustrating HGF expression (RNA-seq) and chromatin accessibility (ATAC-seq log2FC) in K27M and PRC2 mutant cells. While HGF is significantly upregulated in K27M mutants, ATAC-seq analysis does not indicate corresponding chromatin accessibility changes. This suggests that HGF regulation is independent of chromatin opening and may instead be controlled by transcriptional or enhancer-based mechanisms

Full size image

Pharmacological EZH2 inhibition

To further validate the effects of EZH1/2 knockout, we examined the impact of the small molecule EZH inhibitor EPZ-6438. This compound selectively inhibited K27 methylation within the 1–10µM concentration range without causing off-target toxicity, as shown in Figure S9A, B and Figure S10. MTS assays over 72 h indicated minimal effect on cell proliferation (Figure S9C). Extending the investigation to 21 days, EPZ-6438 significantly reduced proliferation in H3.3K27M mutant cells compared to wild-types (Figure S9D). EZH1/2 knockout cells showed resistance to this inhibition (Figure S9E), confirming the specificity of our gene editing and EPZ-6438’s selectivity.

Western blots assessed neural stem cell protein expression in EZH1/2 knockout and control cells, and cells treated with 10µM EPZ-6438 for 72 h. Analyzed proteins included SOX2, Olig2, Nestin, Nanog, GFAP, H3K27M, H3K27Me3, histone H3, and tubulin (Figure S8A). Significant changes in protein profiles suggested EZH2 enzymatic influence on neural stem cell networks, with notable adaptations in Olig2, Nestin, Nanog, and GFAP occurring after EZH1/2 deletion but not after pharmacological inhibition, indicating longer-term effects.

Hepatocyte growth factor (HGF) is known for its role in promoting cell proliferation, migration, and angiogenesis in glioma [57]. In our study, we had observed a marked variation in HGF expression across different cell lines carrying the K27M mutation or EZH1/2 knockouts. Western blot analysis revealed that HGF expression was notably high in SF8628 K27M mutant cells, significantly reduced in wild-type revertants, and absent in EZH1/2 knockout cells (Figure S8B). This pattern indicates that HGF expression is closely associated with the K27M mutation rather than the loss of EZH1/2 or depletion in H3K27me3 levels. Furthermore, treatment with the EZH2 inhibitor EPZ6438 in SF8628 cells led to a partial reduction in HGF expression (Figure S8C), suggesting that EZH2 activity contributes to of HGF regulation but is not its sole determinant. RNA-seq further confirmed these observations, with a significant increase in HGF expression (log2FC = 5.5) in SF8628 K27M cells compared to wild-type controls, (Fig. 7B), whereas DIPG K27M and SF9427 EZH1/2 knockout cells displayed decreased HGF expression (log2FC = -4.5 and − 3.9, respectively) (Fig. 6). Western blot analyses were conducted to investigate HGF expression dysregulation in H3.3K27M mutants and controls, including cells treated with 10µM EPZ-6438 for 72 h and untreated EZH1/2 knockout and control cells (Figure S8B, C). These studies confirmed HGF regulation by the K27M mutation, independent of H3 K27me3 loss.

In vivo evaluation of tumor growth and progression in xenograft model

To evaluate the tumor-initiating capabilities of genetically modified cell lines, immunodeficient nude mice were inoculated with 1 × 10^6 cells from various H3.3 and EZH1/2 CRISPR-modified lines including SF8628 H3.3K27M, SF8628 H3.3 WT, SF8628 H3.3K27M EZH1/2 knockout, SF8628 H3.3 WT EZH1/2 knockout, SF9247 H3.3 WT, and SF9247 EZH1/2 knockout clones. CRSIPR clones of the same type were pooled for engraftment as were examining for consistant loss of fuction/growth. Among these, only SF8628 H3.3K27M cells initiated tumor growth (Figure S11), with other lines failing to form tumors. However, by 360 days, insufficient tumor-bearing mice reached the experimental endpoint to declare a significant difference in overall survival. Further studies explored the effects of CRISPR-induced EZH1/2 deletions on tumor growth using adult glioma (glioblastoma multiforme (GBM)) xenograft models, U87 and U118. Both EZH1/2 knockout (3 pooled clones for engraftment) and control variants of these lines were tested, showing that EZH1/2 knockout significantly delayed tumor onset and increased survival, without changing the growth rate of the tumors in vivo (Figures S12 and S13).

Discussion

Contrary to the prevailing understanding, the precise mechanism by which the H3K27M mutation drives tumorigenesis is incomplete. While it has been widely accepted that the global loss of H3K27me3 in K27M-mutant cells is solely responsible for the associated oncogenic transformation, emerging evidence suggests a more nuanced picture. H3K27M mutation not only induces widespread loss of H3K27me3 but also affects adjacent histone modifications, pointing to a multifaceted impact on the chromatin landscape and gene regulation [34, 58]. Competing molecular models, each supported by experimental evidence, have been proposed to explain how the K27M mutation leads to the depletion of H3K27me3. Competing theories propose that H3.3K27M mutation involves either the sequestration of PRC2 [24,25,26, 59], selective retention of PRC2 activity at specific sites [29], exclusion of PRC2 from certain chromatin regions [30], an allosteric modification (poison model) that impacts PRC2’s distribution and function across the genome [31], or a ‘step down’ gradation from me3 to me2 and me1 [32], suggesting complex interactions that disrupt traditional methylation patterns. While our findings do not directly address this ongoing debate, the preponderance of evidence supports the sequestration model. Nevertheless, our results add a new layer of complexity by suggesting that K27M itself exerts a distinct epigenetic function critical for tumor formation. This function likely acts in synergy with H3K27me3 depletion to drive the transformation of glial cells.

Research on small molecule inhibitors of EZH2 has demonstrated potential for reducing H3K27me3 [29, 30, 60, 61], paradoxically leading to decreased methylation at silenced tumor-suppressor genes in DMG and inhibiting cell proliferation [29, 30]. Our findings corroborate this approach, as both PRC2 knockout mice and K27M revertants showed no tumor development, highlighting the critical role of precise methylation balance “just right model” in tumorigenesis. Deviations from this balance reduce tumorigenic potential, offering therapeutic opportunities. In our study, PRC2 knockout in SF8628 cells prevented tumor formation in mice, in contrast to SF8628 K27M cells, which formed tumors. This supports Mohammad et al.‘s findings [29], on the essential role of PRC2 in tumor proliferation in H3K27M-expressing tumors.

Lewis et al., 2022 [35], report that the H3.3K27M mutation in DIPG lines SU-DIPG-XIII and SU-DIPG-XVII, and their gene-edited wild-type counterparts, leads to open chromatin that enhances expression of crucial developmental genes like ASCL1 and NEUROD1, potentially accelerating tumor progression. Their findings emphasize the mutation’s impact on chromatin dynamics and gene regulation, notably enriching transcription factors such as ASCL1, NEUROD1, OLIG2, and HOXA2 in K27M mutants. Similarly, in our study, genes uniquely dysregulated by K27M exhibited significant chromatin accessibility changes, where upregulated genes were associated with increased accessibility and downregulated genes with reduced accessibility. This pattern reinforces the idea that K27M directly alters chromatin structure, facilitating the activation of specific gene networks. In contrast, our study noted a significant upregulation of ASCL1 in K27M DIPG XVII cells (log2FC = 11.8) without a parallel increase in SF8628 K27M or SF9427 PR2KO cells. NEUROD1, OLIG2, and HOXA2 also showed increased levels in K27M DIPG XVII but not SF8628 cells. These observations suggest that the K27M mutation’s effects on gene expression and chromatin dynamics are cell-type specific and heavily influenced by the cellular context (Fig. 7b). Furthermore, while K27M uniquely dysregulated genes showed a strong association with chromatin accessibility, genes uniquely dysregulated in PRC2KO cells did not exhibit significant differences in chromatin accessibility between upregulated, downregulated, and unchanged genes. This suggests that PRC2 loss influences transcription through mechanisms independent of chromatin remodeling.

In our study, we observed significant upregulation of the CDKN2A gene in K27M mutant models such as SF8628 K27M, SF8628 K27M EZH1/2 KO, and DIPG XVII K27M, with respective log2 fold changes of 12.9, 13.4, and 10.9. This upregulation was notably absent in EZH1/2 knockout models lacking the K27M mutation, suggesting that K27M exerts its regulatory effects on CDKN2A independently of PRC2 activity. In contrast, our ATAC-seq profiles show that HOX genes, typically regulated by PRC2, remain comparatively inaccessible in K27M cells, and become accessible only upon PRC2 knockout, indicating a strong regulatory control by PRC2 rather than histone modification changes induced by H3K27M. A previous study by Cordero FJ et al. [62] reported that H3.3K27M represses transcription specifically at the p16/INK4A (CDKN2A) locus through DNA methylation rather than PRC2-dependent H3K27me3 deposition. In contrast, our findings demonstrate that H3.3K27M drives widespread transcriptional changes across the genome, including significant upregulation of CDKN2A in K27M mutant cells. Moreover, their conclusions, drawn primarily from mouse models without human validation, limit the broader applicability of their findings.

Chen et al., [63] observed that the K27M mutation generally decreases H3K27me3 levels, predominantly leading to gene induction with 139 genes upregulated across K27M lines. In contrast, our study finds that K27M not only interacts with PRC2 but also affects gene regulation more evenly, both upregulating and downregulating genes, with a more significant impact on the downregulation of biological pathways, imposing an overall repressive effect on the transcriptome. This observation indicates that previous studies comparing K27M effects solely with wild-type controls may have missed the mutation’s complex role due to overlapping PRC2 impacts on gene regulation. Our results reveal that K27M alters PRC2 function and independently influences gene expression, suggesting a dual mechanism where K27M can both modify and counteract PRC2 activity, leading to varied regulatory effects on the transcriptome.

We observed that the reversion to wildtype in H3.3K27M mutant cells led to significant reductions in growth and survival over 21 days. The well-documented response of K27M mutant cells to EZH2 inhibition was confirmed in vivo; however, this response was reversed in K27M wildtype revertants, and EZH2 knockouts were found to be insensitive to EZH1/2 inhibition with EPZ-6438. Notably, both K27M wildtype revertants and EZH1/2 knockout cells failed to grow in immunodeficient mice, unlike K27M mutant cells. Furthermore, the glioblastoma cell lines U87 and U118 exhibited a significant reduction or delay in tumor growth in immunodeficient mice following EZH2 deletion. While initially appearing unconnected, these findings provide insights into the effects of PRC2 loss compared to H3K27 methylation inhibition, highlighting its broader relevance to glioma biology. The consistency of these effects across multiple glioma models has implications for both DMG and GBM tumor progression.

The differential responses observed between H3.3K27M, H3.3 wildtype revertants, and EZH2 knockout cells emphasizes the potential specificity and intimal effectiveness of EZH2-targeted therapies in K27M-mutated scenarios. Moreover, the inability of H3.3 WT revertant and EZH1/2 knockout cell lines to proliferate in immunodeficient mice suggests that the K27M mutation may confer unique growth advantages independent from the loss of H3K27me3.

In summary, our findings highlight the K27M mutation’s multifaceted impact on gene expression, chromatin accessibility, and tumor dynamics, which transcends mere reductions in H3K27me3. The mutation introduces a dual regulatory mechanism—modulating and counteracting PRC2—that significantly influences gene expression and biological pathways. Notably, this is the first evidence of genes uniquely deregulated by K27M, independent of K27 methylation loss, show marked changes in chromatin accessibility. This specific pattern, distinct from changes driven by EZH1/2 knockouts, defines H3.3K27M’s unique effect on chromosomal accessibility, independent of its impact on chromatin via loss of K27 methylation.

Data availability

The RNA-seq and ATAC-seq datasets generated and analyzed in this study has been deposited in the NCBI Gene Expression Omnibus (GEO) repository under the accession numbers GSE293770 and GSE293771 respectively.

References

  1. Mackay A, Burford A, Carvalho D, Izquierdo E, Fazal-Salom J, Taylor KR, Bjerke L, Clarke M, Vinci M, Nandhabalan M, Temelso S, Popov S, Molinari V, Raman P, Waanders AJ, Han HJ, Gupta S, Marshall L, Zacharoulis S, et al. Integrated molecular Meta-Analysis of 1,000 pediatric High-Grade and diffuse intrinsic Pontine glioma. Cancer Cell. 2017;32:520–e5375.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Adhikari S, Bhutada AS, Ladner L, Cuoco JA, Entwistle JJ, Marvin EA, Rogers CM. Prognostic Indicators for H3K27M-Mutant Diffuse Midline Glioma: A Population-Based Retrospective Surveillance, Epidemiology, and End Results Database Analysis. World Neurosurg [Internet]. Elsevier Inc.; 2023;178:e113–21. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.wneu.2023.07.001

  3. Sturm D, Witt H, Hovestadt V, Khuong-Quang DA, Jones DT, Konermann C, et al. Hotspot mutations in H3F3A and IDH1 define distinct epigenetic and biological subgroups of glioblastoma. Cancer Cell. 2012;22(4):425–37. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.ccr.2012.08.024. PMID: 23079654.

  4. Khuong-Quang DA, Buczkowicz P, Rakopoulos P, Liu XY, Fontebasso AM, Bouffet E, Bartels U, Albrecht S, Schwartzentruber J, Letourneau L, Bourgey M, Bourque G, Montpetit A, Bourret G, Lepage P, Fleming A, Lichter P, Kool M, Von Deimling A, et al. K27M mutation in histone H3.3 defines clinically and biologically distinct subgroups of pediatric diffuse intrinsic Pontine gliomas. Acta Neuropathol. 2012;124:439–47.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Louis DN, Perry A, Reifenberger G, von Deimling A, Figarella-Branger D, Cavenee WK, Ohgaki H, Wiestler OD, Kleihues P, Ellison DW. The 2016 world health organization classification of tumors of the central nervous system: a summary. Acta Neuropathol. Volume 131. Berlin Heidelberg: Springer; 2016. pp. 803–20.

    Google Scholar 

  6. Yuen BT, Knoepfler PS. Histone H3.3 mutations: a variant path to cancer. Cancer Cell. 2013;24(5):567–74. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.ccr.2013.09.015. PMID: 24229707; PMCID: PMC3882088.

  7. Schwartzentruber J, Korshunov A, Liu XY, Jones DT, Pfaff E, Jacob K, et al. Driver mutations in histone H3.3 and chromatin remodelling genes in paediatric glioblastoma. Nature. 2012;482(7384):226–31. https://doiorg.publicaciones.saludcastillayleon.es/10.1038/nature10833. Erratum in: Nature. 2012;484(7392):130. PMID: 22286061.

  8. Wu G, Broniscer A, McEachron TA, Lu C, Paugh BS, Becksfort J, et al. Somatic histone H3 alterations in pediatric diffuse intrinsic pontine gliomas and non-brainstem glioblastomas. Nat Genet. 2012;44(3):251–3. https://doiorg.publicaciones.saludcastillayleon.es/10.1038/ng.1102. PMID: 22286216; PMCID: PMC3288377.

  9. Montgomery ND, Yee D, Chen A, Kalantry S, Chamberlain SJ, Otte AP, Magnuson T. The murine polycomb group protein Eed is required for global histone H3 lysine-27 methylation. Curr Biol [Internet]. Curr Biol; 2005 [cited 2024 Apr 24];15:942–7. Available from: https://pubmed.ncbi.nlm.nih.gov/15916951/

  10. Shen X, Liu Y, Hsu YJ, Fujiwara Y, Kim J, Mao X, Yuan GC, Orkin SH. EZH1 mediates methylation on histone H3 lysine 27 and complements EZH2 in maintaining stem cell identity and executing pluripotency. Mol Cell. 2008;32.

  11. Cao R, Zhang Y. SUZ12 is required for both the histone methyltransferase activity and the silencing function of the EED-EZH2 complex. Mol Cell [Internet]. Mol Cell; 2004 [cited 2024 Apr 24];15:57–67. Available from: https://pubmed.ncbi.nlm.nih.gov/15225548/

  12. Margueron R, Li G, Sarma K, Blais A, Zavadil J, Woodcock CL, Dynlacht BD, Reinberg D. Ezh1 and Ezh2 Maintain Repressive Chromatin through Different Mechanisms. Mol Cell [Internet]. Elsevier Inc.; 2008;32:503–18. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.molcel.2008.11.004

  13. Shan Y, Liang Z, Xing Q, Zhang T, Wang B, Tian S, Huang W, Zhang Y, Yao J, Zhu Y, Huang K, Liu Y, Wang X, Chen Q, Zhang J, Shang B, Li S, Shi X, Liao B, et al. PRC2 specifies ectoderm lineages and maintains pluripotency in primed but not Naïve ESCs. Nat Commun. 2017;8.

  14. Boyer LA, Plath K, Zeitlinger J, Brambrink T, Medeiros LA, Lee TI, Levine SS, Wernig M, Tajonar A, Ray MK, Bell GW, Otte AP, Vidal M, Gifford DK, Young RA, Jaenisch R. Polycomb complexes repress developmental regulators in murine embryonic stem cells. Nat 2006 4417091 [Internet]. Nature Publishing Group; 2006 [cited 2024 Apr 24];441:349–53. Available from: https://www.nature.com/articles/nature04733

  15. Di Croce L, Helin K. Transcriptional regulation by Polycomb group proteins. Nat Struct Mol Biol [Internet]. Nat Struct Mol Biol; 2013 [cited 2024 Apr 24];20:1147–55. Available from: https://pubmed.ncbi.nlm.nih.gov/24096405/

  16. Schuettengruber B, Cavalli G. Recruitment of polycomb group complexes and their role in the dynamic regulation of cell fate choice. Development [Internet]. Development; 2009 [cited 2024 Apr 24];136:3531–42. Available from: https://pubmed.ncbi.nlm.nih.gov/19820181/

  17. Kim KH, Roberts CW. Targeting EZH2 in cancer. Nat Med. 2016;22(2):128–34. https://doiorg.publicaciones.saludcastillayleon.es/10.1038/nm.4036. PMID: 26845405; PMCID: PMC4918227.

  18. Lee W, Teckie S, Wiesner T, Ran L, Prieto Granada CN, Lin M, Zhu S, Cao Z, Liang Y, Sboner A, Tap WD, Fletcher JA, Huberman KH, Qin LX, Viale A, Singer S, Zheng D, Berger MF, Chen Y, et al. PRC2 is recurrently inactivated through EED or SUZ12 loss in malignant peripheral nerve sheath tumors. Nat Genet [Internet]. Nat Genet; 2014 [cited 2024 Apr 24];46:1227–32. Available from: https://pubmed.ncbi.nlm.nih.gov/25240281/

  19. McCabe MT, Graves AP, Ganji G, Diaz E, Halsey WS, Jiang Y, Smitheman KN, Ott HM, Pappalardi MB, Allen KE, Chen SB, Pietra A, Della, Dul E, Hughes AM, Gilbert SA, Thrall SH, Tummino PJ, Kruger RG, Brandt M, et al. Mutation of A677 in histone methyltransferase EZH2 in human B-cell lymphoma promotes hypertrimethylation of histone H3 on lysine 27 (H3K27). Proc Natl Acad Sci U S A [Internet]. Proc Natl Acad Sci U S A; 2012 [cited 2024 Apr 24];109:2989–94. Available from: https://pubmed.ncbi.nlm.nih.gov/22323599/

  20. Morin RD, Johnson NA, Severson TM, Mungall AJ, An J, Goya R, Paul JE, Boyle M, Woolcock BW, Kuchenbauer F, Yap D, Humphries RK, Griffith OL, Shah S, Zhu H, Kimbara M, Shashkin P, Charlot JF, Tcherpakov M, et al. Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nat Genet 2010 422 [Internet]. Nature Publishing Group; 2010 [cited 2024 Apr 24];42:181–5. Available from: https://www.nature.com/articles/ng.518

  21. Bachmann IM, Halvorsen OJ, Collett K, Stefansson IM, Straume O, Haukaas SA, Salvesen HB, Otte AP, Akslen LA. EZH2 expression is associated with high proliferation rate and aggressive tumor subgroups in cutaneous melanoma and cancers of the endometrium, prostate, and breast. J Clin Oncol [Internet]. J Clin Oncol; 2006 [cited 2024 Apr 24];24:268–73. Available from: https://pubmed.ncbi.nlm.nih.gov/16330673/

  22. Varambally S, Dhanasekaran SM, Zhou M, Barrette TR, Kumar-Sinha C, Sanda MG, Ghosh D, Pienta KJ, Sewalt RGAB, Rubin MA, Chinnaiyan AM. The polycomb group protein EZH2 is involved in progression of prostate cancer. Nature [Internet]. Nature; 2002 [cited 2024 Apr 24];419:624–9. Available from: https://pubmed.ncbi.nlm.nih.gov/12374981/

  23. Zhang M, Wang Y, Jones S, Sausen M, McMahon K, Sharma R, Wang Q, Belzberg AJ, Chaichana K, Gallia GL, Gokaslan ZL, Riggins GJ, Wolinksy JP, Wood LD, Montgomery EA, Hruban RH, Kinzler KW, Papadopoulos N, Vogelstein B, et al. Somatic mutations of SUZ12 in malignant peripheral nerve sheath tumors. Nat Genet [Internet]. Nat Genet; 2014 [cited 2024 Apr 24];46:1170–2. Available from: https://pubmed.ncbi.nlm.nih.gov/25305755/

  24. Fang D, Gan H, Cheng L, Lee JH, Zhou H, Sarkaria JN, et al. H3.3K27M mutant proteins reprogram epigenome by sequestering the PRC2 complex to poised enhancers. Elife. 2018;7:e36696. https://doiorg.publicaciones.saludcastillayleon.es/10.7554/eLife.36696. PMID: 29932419; PMCID: PMC6033537.

  25. Diehl KL, Ge EJ, Weinberg DN, Jani KS, Allis CD, Muir TW. PRC2 engages a bivalent H3K27M-H3K27me3 dinucleosome inhibitor. Proc Natl Acad Sci U S United States. 2019;116:22152–7.

    Article  CAS  Google Scholar 

  26. Justin N, Zhang Y, Tarricone C, Martin SR, Chen S, Underwood E, De Marco V, Haire LF, Walker PA, Reinberg D, Wilson JR, Gamblin SJ. Structural basis of oncogenic histone H3K27M inhibition of human polycomb repressive complex 2. Nat Commun 2016 71 [Internet]. Nature Publishing Group; 2016 [cited 2023 Dec 13];7:1–11. Available from: https://www.nature.com/articles/ncomms11316

  27. Lewis PW, Müller MM, Koletsky MS, Cordero F, Lin S, Banaszynski LA, Garcia BA, Muir TW, Becher OJ, Allis CD. Inhibition of PRC2 activity by a gain-of-function H3 mutation found in pediatric glioblastoma. Science (80-) [Internet]. American Association for the Advancement of Science; 2013 [cited 2023 Dec 11];340:857–61. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1126/science.1232245

  28. Bender S, Tang Y, Lindroth AM, Hovestadt V, Jones DTW, Kool M, Zapatka M, Northcott PA, Sturm D, Wang W, Radlwimmer B, Højfeldt JW, Truffaux N, Castel D, Schubert S, Ryzhova M, Şeker-Cin H, Gronych J, Johann PD et al. Reduced H3K27me3 and DNA Hypomethylation Are Major Drivers of Gene Expression in K27M Mutant Pediatric High-Grade Gliomas. Cancer Cell [Internet]. Cell Press; 2013 [cited 2023 Dec 11];24:660–72. Available from: http://www.cell.com/article/S1535610813004558/fulltext

  29. Mohammad F, Weissmann S, Leblanc B, Pandey DP, Højfeldt JW, Comet I, Zheng C, Johansen JV, Rapin N, Porse BT, Tvardovskiy A, Jensen ON, Olaciregui NG, Lavarino C, Suñol M, De Torres C, Mora J, Carcaboso AM, Helin K. EZH2 is a potential therapeutic target for H3K27M-mutant pediatric gliomas. Nat Med Nat Publishing Group. 2017;23:483–92.

    CAS  Google Scholar 

  30. Piunti A, Hashizume R, Morgan MA, Bartom ET, Horbinski CM, Marshall SA, Rendleman EJ, Ma Q, Takahashi YH, Woodfin AR, Misharin AV, Abshiru NA, Lulla RR, Saratsis AM, Kelleher NL, James CD, Shilatifard A. Therapeutic targeting of polycomb and BET bromodomain proteins in diffuse intrinsic Pontine gliomas. Nat Med Nat Publishing Group. 2017;23:493–500.

    CAS  Google Scholar 

  31. Stafford JM, Lee CH, Voigt P, Descostes N, Saldana-Meyer R, Yu JR, Leroy G, Oksuz O, Chapman JR, Suarez F, Modrek AS, Bayin NS, Placantonakis DG, Karajannis MA, Snuderl M, Ueberheide B, Reinberg D. Multiple modes of PRC2 inhibition elicit global chromatin alterations in H3K27M pediatric glioma. Sci Adv [Internet]. American Association for the Advancement of Science; 2018 [cited 2023 Dec 13];4:5935–66. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1126/sciadv.aau5935

  32. Harutyunyan AS, Chen H, Lu T, Horth C, Nikbakht H, Krug B, Russo C, Bareke E, Marchione DM, Coradin M, Garcia BA, Jabado N, Majewski J. H3K27M in Gliomas Causes a One-Step Decrease in H3K27 Methylation and Reduced Spreading within the Constraints of H3K36 Methylation. Cell Rep [Internet]. United States; 2020;33:108390. Available from: http://www.ncbi.nlm.nih.gov/pubmed/33207202

  33. Cohen LRZ, Kaffe B, Deri E, Leibson C, Nissim-Rafinia M, Maman M, Harpaz N, Ron G, Shema E, Meshorer E. PRC2-independent actions of H3.3K27M in embryonic stem cell differentiation. Nucleic Acids Res. Oxford University Press; 2023;51:1662–73.

  34. Day CA, Grigore F, Hakkim FL, Daniels DJ, Robinson JP, Hinchcliffe EH. The histone H3. 3 K27M mutation suppresses Ser31phosphorylation and mitotic fidelity, which can directly drive gliomagenesis. Curr Biol [Internet]. Elsevier Inc.; 2025;35:1–9. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.cub.2024.11.035

  35. Lewis NA, Klein RH, Kelly C, Yee J, Knoepfler PS, Histone. H3.3 K27M chromatin functions implicate a network of neurodevelopmental factors including ASCL1 and NEUROD1 in DIPG. Epigenetics and Chromatin [Internet]. BioMed Central; 2022;15:1–18. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13072-022-00447-6

  36. Hashizume R, Andor N, Ihara Y, Lerner R, Gan H, Chen X, Fang D, Huang X, Tom MW, Ngo V, Solomon D, Mueller S, Paris PL, Zhang Z, Petritsch C, Gupta N, Waldman TA, James CD. Pharmacologic inhibition of histone demethylation as a therapy for pediatric brainstem glioma. Nat Med [Internet]. Nature Publishing Group; 2014;20:1394–6. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1038/nm.3716

  37. Qin EY, Cooper DD, Abbott KL, Lennon J, Nagaraja S, Mackay A, Jones C, Vogel H, Jackson PK, Monje M. Neural Precursor-Derived Pleiotrophin mediates subventricular zone invasion by glioma. Cell Elsevier Inc. 2017;170:845–e85919.

    CAS  Google Scholar 

  38. Nagaraja S, Vitanza NA, Woo PJ, Taylor KR, Liu F, Zhang L, Li M, Meng W, Ponnuswami A, Sun W, Ma J, Hulleman E, Swigut T, Wysocka J, Tang Y, Monje M. Transcriptional Dependencies in Diffuse Intrinsic Pontine Glioma. Cancer Cell [Internet]. Elsevier Inc.; 2017;31:635–652.e6. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.ccell.2017.03.011

  39. Meier JA, Zhang F, Sanjana NE. GUIDES: SgRNA design for loss-of-function screens. Nat Methods. 2017;14:831–2.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Chang K, Marran K, Valentine A, Hannon GJ. Packaging ShRNA retroviruses. Cold Spring Harb Protoc. 2013;2013:734–7.

    Article  PubMed  Google Scholar 

  41. Grigore F, Yang H, Hanson ND, VanBrocklin MW, Sarver AL, Robinson JP. BRAF inhibition in melanoma is associated with the dysregulation of histone methylation and histone methyltransferases. Neoplasia (United States) [Internet]. Elsevier Inc.; 2020;22:376–89. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.neo.2020.06.006

  42. Andrews S, FastQC. A Quality Control Tool for High Throughput Sequence Data [Internet]. 2010. Available from: http://www.bioinformatics.babraham.ac.uk/projects/fastqc/

  43. Chen S, Zhou Y, Chen Y, Gu J. Fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics. 2018;34:i884–90.

    Article  PubMed  PubMed Central  Google Scholar 

  44. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29:15–21.

    Article  CAS  PubMed  Google Scholar 

  45. Liao Y, Smyth GK, Shi W, FeatureCounts. An efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 2014;30:923–30.

  46. Love MI, Huber W, Anders S. Moderated Estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:1–21.

    Article  Google Scholar 

  47. Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, Feng T, Zhou L, Tang W, Zhan L, Fu X, Liu S, Bo X, Yu G. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation [Internet]. Elsevier; 2021;2:100141. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.xinn.2021.100141

  48. Blighe K, Rana SLM, EnhancedVolcano. Publication-ready volcano plots with enhanced colouring and labeling. 2023; Available from: https://bioconductor.org/packages/EnhancedVolcano

  49. Chen H, Boutros PC. VennDiagram: a package for the generation of highly-customizable Venn and Euler diagrams in R. BMC Bioinformatics [Internet]. 2011;12:35. Available from: https://biomedcentral-www.publicaciones.saludcastillayleon.es/1471-2105/12/35

  50. Szklarczyk D, Kirsch R, Koutrouli M, Nastou K, Mehryary F, Hachilif R, Gable AL, Fang T, Doncheva NT, Pyysalo S, Bork P, Jensen LJ, Von Mering C. The STRING database in 2023: protein-protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res. Oxford University Press; 2023;51:D638–46.

  51. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25:1754–60.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  52. Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, Nussbaum C, Myers RM, Brown M, Li W, Shirley XS. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 2008;9.

  53. Kungulovski G, Jeltsch A. Quality of histone modification antibodies undermines chromatin biology research. F1000Research. 2015;4:1160.

    Article  PubMed  PubMed Central  Google Scholar 

  54. Rothbart SB, Dickson BM, Raab JR, Grzybowski AT, Krajewski K, Guo AH, et al. An interactive database for the assessment of histone antibody specificity. Mol Cell. 2015;59(3):502–11. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.molcel.2015.06.022. Epub 2015 Jul 23. PMID: 26212453; PMCID: PMC4530063.

  55. Zhao Y, Jensen ON. Modification-specific proteomics: strategies for characterization of post-translational modifications using enrichment techniques. Proteomics. 2009;9(20):4632–41. https://doiorg.publicaciones.saludcastillayleon.es/10.1002/pmic.200900398. PMID: 19743430; PMCID: PMC2892724.

  56. Lavarone E, Barbieri CM, Pasini D. Dissecting the role of H3K27 acetylation and methylation in PRC2 mediated control of cellular identity. Nat Commun. 2019;10.

  57. Petterson SA, Dahlrot RH, Hermansen SK, Munthe KA, Gundesen S, Wohlleben MT, Rasmussen H, Beier T, Hansen CP, Kristensen S. BW. High levels of c-Met is associated with poor prognosis in glioblastoma. J Neurooncol [Internet]. Springer US; 2015;122:517–27. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1007/s11060-015-1723-3

  58. Brien GL, Bressan RB, Monger C, Gannon D, Lagan E, Doherty AM, Healy E, Neikes H, Fitzpatrick DJ, Deevy O, Grant V, Marqués-Torrejón M-AA, Alfazema N, Pollard SM, Bracken AP. Simultaneous disruption of PRC2 and enhancer function underlies histone H3.3-K27M oncogenic activity in human hindbrain neural stem cells. Nat Genet [Internet]. United States: Springer US; 2021;53:1221–32. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1038/s41588-021-00897-w

  59. Lewis PW, MM M, Francisco SKM, Shu C, A BL L, GB A, MT W, Allis JBO. David. Inhibition of PRC2 activity by a Gain-of-Function H3 mutation found in pedriatic glioblastoma. Science. 2011;23:1–7. (80-).

    Google Scholar 

  60. Hennika T, Hu G, Olaciregui NG, Barton KL, Ehteda A, Chitranjan A, Chang C, Gifford AJ, Tsoli M, Ziegler DS, Carcaboso AM, Becher OJ. Pre-clinical study of Panobinostat in xenograft and genetically engineered murine diffuse intrinsic Pontine glioma models. PLoS ONE. 2017;12:1–20.

    Article  Google Scholar 

  61. Grasso CS, Tang Y, Truffaux N, Berlow NE, Liu L, Debily MA, Quist MJ, Davis LE, Huang EC, Woo PJ, Ponnuswami A, Chen S, Johung TB, Sun W, Kogiso M, Du Y, Qi L, Huang Y, Hütt-Cabezas M, et al. Functionally defined therapeutic targets in diffuse intrinsic Pontine glioma. Nat Med. 2015;21:555–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  62. Cordero FJ, Huang Z, Grenier C, He X, Hu G, McLendon RE, Murphy SK, Hashizume R, Becher OJ. Histone H3.3K27M represses p16 to accelerate gliomagenesis in a murine model of DIPG. Mol Cancer Res. 2017;15:1243–54.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  63. Chen KY, Bush K, Klein RH, Cervantes V, Lewis N, Naqvi A, Carcaboso AM, Lechpammer M, Knoepfler PS. Reciprocal H3.3 gene editing identifies K27M and G34R mechanisms in pediatric glioma including NOTCH signaling. Commun Biol. 2020;3.

Download references

Acknowledgements

We thank Paul Knoepfler (UC Davis School of Medicine, Sacramento, CA) for kindly providing the DIPGXVII DMG H3.3 K27M mutant and wild-type revertant cells.

Funding

This research was funded by the US Department of Defense Army under Grant Numbers W81XWH2211045-P00001 (PI: JPR, University of Minnesota), W81XWH-21-1-0546 (PI: EHH, University of Minnesota), and W81XWH-18-1-0493 (PI: EHH, University of Minnesota). Additional support was provided by the Minnesota Partnership for Biotechnology and Medical Genomics Collaborative Research Grant (PIs: EHH and JPR) and R01HL125353 from National Institute of Health, NHLBI (Co-PI: EHH), and T32 Training Grant, T32CA217836, Mayo Clinic (CAD).

Author information

Author notes

  1. Suraj Bhattarai, Faruck L. Hakkim and Charles A. Day are joint first authors.

Authors and Affiliations

  1. The Hormel Institute, University of Minnesota, 801 16th Avenue NE, Austin, MN, 55912, USA

    Suraj Bhattarai, Faruck L. Hakkim, Charles A. Day, Alyssa Langfald, Igor Entin, Edward H. Hinchcliffe & James P. Robinson

  2. Neuro-Oncology Training Program, Mayo Clinic, Rochester, MN, USA

    Charles A. Day

  3. Department of Neurosurgery, University of Minnesota, Minneapolis, MN, USA

    Florina Grigore

  4. Masonic Cancer Center, University of Minnesota, Minneapolis, MN, USA

    Edward H. Hinchcliffe & James P. Robinson

Authors
  1. Suraj Bhattarai
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Faruck L. Hakkim
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Charles A. Day
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Florina Grigore
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Alyssa Langfald
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Igor Entin
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Edward H. Hinchcliffe
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. James P. Robinson
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

S.B analyzed data and wrote the manuscript. F.L.H and C.A.D performed the study. F.G assisted with data analysis. A.L and I.E supported the research. EHH supervised C.A.D and AL. J.P.R, as Principal Investigator, designed the study, supervised the research, and wrote the manuscript.

Corresponding author

Correspondence to James P. Robinson.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary Material 2

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Bhattarai, S., Hakkim, F.L., Day, C.A. et al. H3F3A K27M mutations drive a repressive transcriptome by modulating chromatin accessibility independent of H3K27me3 in Diffuse Midline Glioma. Epigenetics & Chromatin 18, 23 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13072-025-00585-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13072-025-00585-7

Keywords

Epigenetics & Chromatin

ISSN: 1756-8935