Onco-histone mutations in diffuse intrinsic pontine glioma

Abstract

Diffuse intrinsic pontine glioma (DIPG) is a deadly pediatric tumor with no available treatment options. Over 60–70% DIPG tumors harbor heterozygous mutations at genes encoding histone H3 proteins that replace lysine 27 with methionine. In this review, we discuss how K27M mutation reprograms the cancer epigenome leading to tumorigenesis, and potential drug targets and therapeutic agents for DIPG.

Diffuse intrinsic pontine gliomas – an overview

Diffuse intrinsic pontine glioma (DIPG) is a rare tumor that affects approximately 200–300 children in the United States per year, but represents the greatest cause of brain tumor-related deaths under the age of 19 [1–5]. Most children diagnosed with DIPG are 6 to 8 years old, and the medium survival for these patients is less than one year after diagnosis [6]. While significant progress has been made in treating of other pediatric tumors, no improvement in survival has been achieved in more than four decades in DIPG patients. Because of its precarious location at pons that contain nerves critical for life-sustaining functions, as well as its infiltrative nature, resection of DIPG tumors is not an option. Currently, only radiation shows limited benefits. Moreover, most previous clinical trials on DIPG tumors were based on information gathered from adult glioblastoma [7, 8]. Recent genomic, epigenomic and transcriptomic characterization of DIPG tumors indicate that these tumors represent a completely different subgroup of high-grade gliomas with unique mutational and gene expression signatures [6, 9, 10]. Therefore, there is much interest in understanding the tumorigenesis mechanisms of DIPG and to identify novel therapeutic agents. In this review, we will summarize recent advances on the impacts of histone mutations on cancer epigenomes and discuss potential therapeutic agents for DIPG tumors. Other recent reviews comprehensively describe genomic alterations and mutations of DIPG and other high-grade gliomas [2, 3].

Mutations in histone genes are driver mutations for DIPG

Two groups first reported the surprising discovery of somatic mutations on histone genes in hGG tumors [11, 12]. Wu et al performed whole-genome sequencing of 7 DIPG tumors and targeted sequencing of additional 43 DIPG tumors and 36 non-brainstem pediatric glioblastomas and found that 78% DIPG tumors and 22% non-brainstem pediatric glioblastomas contain heterozygous mutations in H3F3A or HIST1H3B [12]. H3F3A is one of two genes encoding the histone H3 variant H3.3, whereas HIST1H3B is one of 13 genes encoding canonical histone H3 (H3.1 and H3.2) (Box 1). H3.3 and H3.1 differ by five amino acids and are assembled into nucleosomes via distinct pathways [13, 14]. In all DIPG cases, mutations in H3F3A and HIST1H3B lead to the replacement of lysine 27 with methionine (K27M) in histone H3. An independent study of exosome sequencing in 48 pediatric glioblastomas found that somatic mutations of histone genes including H3.3K27M and H3.3G34R/V occur in 31% cases [11]. Subsequent analysis of large cohort of high grade gliomas (hGG) showed that mutations in H3K27M and H3.3G34R define two distinct classes of pediatric GBM, with H3K27M tumors found in middle line and pones, and H3.3G34R/V tumors restricted to the cerebral hemisphere [6, 9, 15, 16]. The large cohort studies also indicate that mutations at the H3F3A and HIST1H3B account for 85% and 15% of all K27M tumors, respectively. As discussed below, while the expression of H3.1K27M and H3.3K27M mutant proteins leads to a global reduction in H3K27 methylation, H3.1K27M and H3.3K27M define two sub-groups of DIPG tumors [6, 17]. For instance, H3.3K27M DIPG tumors have a proneural phenotype and a pro-metastatic gene expression signature, whereas H3.1K27M DIPG have a mesenchymal gene signature [6, 10]. Moreover, a large fraction of H3.1K27M, but not H3.3K27M tumors also harbor mutations in ACVR1 [15, 16], which encodes a protein involved in bone morphogenic (BMP) pathway. Recently, it has been shown that the ACVR1 mutation cooperates with H3.1K27M to promote mesenchymal profile in DIPG tumors in part through the Stat3 activation [18]. In the future, it would be important to determine to what extent mutant H3.1K27M and H3.3K27M impact the cancer epigenome differently and whether these differences can be exploited for tailored therapy of H3.1K27M and H3.3K27M DIPG tumors.

Box 1: histone H3.1/3.3 and nucleosome assembly.

In human cells, two genes, H3F3A and H3F3B, encode histone H3 variant H3.3. In contrast, 13 genes encode canonical histone H3 variants including H3.1/H3.2. At the amino acid levels, H3.1/H3.2 and H3.3 differ by four or five amino acids. Moreover, the expression of H3.1/H3.2 peaks during S phase of the cell cycle, whereas H3.3 is expressed during all phase of the cell cycle [60]. In addition to these differences, H3.1/H3.2 and H3.3 are assembled in nucleosomes via different pathways (reviewed in [13, 61]). H3.1/H3.2 are incorporated into chromatin during DNA replication and bind specifically to histone chaperone, Chromatin Assembly Factor I (CAF-1)[60, 62]. It is believed that CAF-1 is recruited to DNA replication forks through its interaction with Proliferating Cell Nuclear Antigen (PCNA), a DNA polymerase clamp [63]. H3.3 deposition is replication-independent and is mediated by histone chaperones, HIRA complex and death domain-associated protein 6 (DAXX) [64–66]. In addition, chromatin remodelers including CHD1 and CHD2 are also involved in this process [67]. The amino acids unique to H3.1/H3.2 and H3.3 are important mediators for the interaction between H3.1/H3.2 and H3.3 with their corresponding histone chaperones and thereby for the nucleosome assembly pathway choices.

Impact of histone mutations on cancer epigenomes

In DIPG tumors with the H3K27M mutation and hGGs with an H3.3G34R/V mutation, mutated histone genes appear not to be amplified. As the mutant proteins are expressed from a mutant allele of 15 different genes encoding histone H3 proteins, the mutant proteins should be very low compared to wild type histone H3. Indeed, it was estimated that the amount of H3.3K27M mutant proteins in DIPG PDX lines is 3–17% of total histone H3 [19–22]. Thus, the H3K27M mutation, likely other onco-histone mutations such as H3.3G34R/V, has a dominant role in tumorigenesis.

Histone H3 lysine 27 is an important residue as its modifications are associated with both gene silencing and active gene transcription. H3K27 can be mono-, di- and tri-methylated (H3K27me1-me3) by the Polycomb Repressive Complex 2 (PRC2) with the Ezh1/2 as the catalytic subunit [23, 24] (Box 2). PRC2, discovered for its role as a transcriptional repressor in Drosophila, is conserved in higher eukaryotic cells. The PRC2 complex and H3K27me3 play an important role in gene silencing during development. In contrast, H3K27 acetylation (H3K27ac), catalyzed by p300/CBP (a lysine acetyltransferase), marks regulatory elements, such as promoters and enhancers of actively transcribed genes [25]. Early studies indicated that DIPG cells with low levels of H3K27M proteins have reduced H3K27 methylation [21, 22, 26]. The low levels of H3K27 methylation is due to the expression of H3K27M mutant proteins as the expression of a H3K27M (including H3.1K27M and H3.3K27M), but not H3K27R mutant transgene, resulted in a global reduction of H3K27 methylation in a cell type/organism independent manner. Remarkably, expression of histone mutant transgenes replacing other lysine residues including H3K4, H3K9 and H3K36 with methionine also leads to a reduction of the methylation of the corresponding histone lysine residue in cells, suggesting a common mechanism for the impact of histone K to M mutant transgenes on histone methylation [21, 27]. Moreover, H3K36M mutation have been found in chondroblastoma [28] and head and neck cancer[29], which drives the reduction of H3K36 methylation in these tumors [30, 31]. In contrast to the reduction of H3K27 methylation in DIPG cells, the levels of H3K27ac increase in these cells [21, 32]. Recently, it has been shown that H3K27ac increases at repetitive DNA sequences including endogenous retrovirus (ERV) [33]. Furthermore, H3K36 methylation was also altered in DIPG cells [34], likely due to the cross talk between H3K27 methylation and H3K36 methylation[35, 36]. In addition to histone modifications, it has been reported that DNA methylation is reduced in DIPG cells [26]. Thus, epigenomes in DIPG cells are altered in several distinct ways (Figure 1A). The initial event driving all these changes is likely the global reduction of H3K27 methylation.

Box2: PRC2 complex and H3K27 methylation mediated silencing.

The Polycomb repressive complexes 1 and 2 (PRC1 and PRC2, respectively) play an important role in silencing gene transcription. In mammalian cells, PRC2 consists of four core subunits, EZH1/2 (enhancer of zeste 1 or 2), EED (embryonic ectoderm development), SUZ12 (suppressor of zeste 12) and RbAp48 (retinoblastoma-binding protein p48; also known as RBAP4) (reviewed in [68]). The signature activity of PRC2 is to methylate histone H3 of Lys27 (H3K27), which is catalyzed by the SET domain of EZH1/2[69, 70]. EZH1/2 alone has minimal activities, but shows strong activity when assembled into complexes with EED and SUZ12 [71, 72]. EED has the ability to bind H3K27me3 and allosterically enhances the methyltransferase activity of PRC2 complex [73]. In Drosophila, the PRC2 complex is recruited to target genomic loci for histone H3K27 methylation through DNA sequence specific elements. In human cells, the PRC2 complex is recruited to specific loci through multiple mechanisms including non-coding RNA and transcription factors (reviewed in [74]). Histone variants are also reported to regulate PRC2 occupancy [75]. H3K27me3 provides a docking site for the PRC1 complex, which contains E3 ubiquitin ligase for histone H2A ubiquitylation. The association of PRC1 with H3K27me3-marked chromatin will induce chromatin compaction, thereby contributing to gene silencing.

Fig. 1.

H3K27M mutation acts as a driver for epigenetic reprogramming and DIPG tumorigenesis.

(A) H3K27M mutant proteins result in substantial epigenetic reprogramming, including global reduction of H3K27me2/3 and DNA methylation, and an increase of H3K27 acetylation and H3K36me3.

(B) In genetically engineered mouse models, H3K27M mutation accelerates tumorigenesis, especially for brainstem HGG and high grade DIPG when combined with Trp53 deletion and activation of the PDGFRA/B.

In contrast to histone K-M mutation, expression of H3G34R/V mutant proteins has little impact on global H3K36 methylation [21, 27]. H3.3G34R/V mutation affects H3K36 methylation in cis [37]. These results indicate that different histone mutations found in cancer cells impact histone modifications differently. An in-depth analysis of histone mutations TCGA database reveals that somatic mutations in histone occur in approximately 4% of a variety of other tumor types, such as bladder cancer, breast cancer and colorectal cancer. These mutations are found on core histones H2A, H2B, and H4, besides H3 [38]. Based on the location of mutations, it is hypothesized that they likely affect the function of the chromatin remodeling SWI/SNF complex, which is mutated in many cancers (reviewed in [39]). Furthermore, some of these mutations may impact the stability of nucleosomes [38, 40]. It will be important to test these predictions and determine how different mutations in onco-histone alter epigenomes. We expect that mechanistic insights gained from the characterization of these histone mutations evolved during tumorigenesis process will not only increase our understanding of the impact of these histone mutations on oncogenic pathways, but also further advance our understandings of epigenomic regulations and plasticity.

Expression of H3K27M leads to reduction in H3K27 methylation in trans

How does expression of mutant H3K27M, at low levels, lead to a global reduction of H3K27 methylation? In vitro, H3K27M peptides and nucleosomes containing H3K27M inhibit the enzymatic activity of PRC2 towards wild type histone and mono-nucleosome substrates [31]. Similarly, H3K36M and H3K9M peptides also inhibit the enzymatic activity of H3K36 methyltransferase and H3K9 methyltransferase, respectively [30, 31, 41]. These results support the idea that K-M mutant proteins likely bind to their cognate enzymes with a higher affinity compared to wild type histones. In vitro, S-adenosine-methionine (SAM), the co-factor for lysine methylation, is needed for the increased binding affinity of the H3K9M to Clr4, the H3K9 methyltransferase in S. pombe [41]. In cells, the PRC2 complexes are enriched for H3.3K27M mutant mononucleosomes compared to wild type H3.3K27M mononucleosomes[22]. These results support the idea that H3K27M mutant proteins likely inhibit the enzymatic activity of the PRC2 complex.

Analysis of the distribution of H3K27M mutant proteins as well as Ezh2 in cells, however, reveals that multiple mechanisms likely account for the reduction of H3K27 methylation in cells. H3K27M mutant proteins, like those of wild type H3.3, are enriched at promoters as well as gene bodies of actively transcribed genes where the PRC2 complex is absent or low [32, 42]. Moreover, the amount of H3K27M mutant proteins are low or absent at strong PRC2 sites with H3K27 methylation [42]. Therefore, it is unlikely that the global reduction of H3K27 methylation is solely due to inhibition of PRC2 complex in cells. Using mouse ES cells engineered with the heterozygous H3.3K27M mutation, we found that Ezh2 were enriched at poised enhancers/weak promoters in H3K27M cells compared to wild type cells[42]. Moreover, H3K27 methylation levels at these sites are reduced in H3K27M mutant cells compared to wild type ES cells despite the presence of Ezh2. These results indicate that Ezh2, and possibly the PRC2 complex, is likely re-distributed to poised enhancers/weak promoters, but unable to methylate wild type histones, by the presence of H3K27M mutant proteins at these genomic loci (Figure 2). Analysis of Ezh2, H3K27M and H3K27me3 ChIP-seq datasets at DIPG lines also identified genomic regions where Ezh2 and H3.3K27M co-localize[42]. These regions are not at actively transcribed regions. Instead, in neuro-precursor cells, these regions are also likely poised enhancers/weak promoters with low levels of H3K27me3. Since the total amount of PRC2 complex appears to be unchanged in H3K27M mutant cells, the redistribution of the PRC2 complex to poised enhancers likely contributes to the global reduction of H3K27 methylation in DIPG cells or in any cells expressing K27M mutant proteins. Consistent with this observation, living cell single molecule image studies indicate that the average residence time of PRC2 complex on chromatin increases in K27M mutant cells [43], likely due to the sequestering the PRC2 complex as poised enhancers. In an independent study, the PRC2 complex purified from the H3.3K27M cells is less active compared to PRC2 complex purified from cells with wild type histones. The reduced activity is likely due to a conformational change of the PRC2 complex induced by H3.3K27M mutant proteins [34]. In the future, it would be interesting to determine how H3K27M mutant proteins induce a conformation change in PRC2. Together, these results indicate that multiple mechanisms contribute to the global reduction of H3K27 methylation in H3K27M mutant cells.

Fig. 2.

A model for the global reduction of H3K27 methylation in H3K27M mutant cells. In wide type cells, most of PRC2 bind to strong PRC sites, which have high levels of H3K27 methylation and contain low level of H3.3. Low levels of PRC2 are also found at poised enhancers that have low levels of H3K27me3. In H3K27M mutant cells, we hypothesize that PRC2 complexes are recruited to poised enhancers via a H3.3K27M independent mechanism and are then trapped there by the presence H3.3K27M proteins, which leads to reduced binding of PRC2 to strong PRC2 sites and subsequent global reduction of H3K27 methylation.

Mechanisms of tumorigenesis in DIPG

Various models were generated to understand how H3.3K27M mutation promotes tumorigenesis. In all experimental models, the H3.3K27M mutant alone is not sufficient to generate tumors (Figure 1B). Instead, it must cooperate with PDGFRA overexpression and/or TP53 loss to efficiently produce tumors. Expression of H3K27M mutant transgenes in human pluripotent stem cells leads to a moderate increase in proliferation. When combined with p53 depletion and overexpression of PDGFRA, triple mutant cells showed marked increase in proliferation. Moreover, tumor form when the triple mutant cells were transplanted into mice [44]. In an independent, it has been shown that H3.3K27M, Trp53 mutation and PDGFRB overexpression also transformed mouse neural stem cells into tumors. Moreover, the gene expression signature of the produced tumor is similar to that of human DIPG cells [45]. DIPG tumors was also generated by combination of H3.3K27M overexpression, Trp53 loss and PDGFB overexpression in neonatal brainstem using in vivo retrovirus transduction [46]. Recently, using genetically engineered mouse models, the impact of H3.3K27M mutation on tumorigenesis was analyzed when it was combined with Trp53 deletion, overexpression of PDGFRA alone or in combinations. It was found that neonatal induction of H3K27M when combined with loss of Trp53 accelerate formation of tumors resembling medulloblastoma. This outcome is different from an early study showing that pre-natal induction of H3K27M and Trp53 depletion led to the generation of DIPG tumors, suggesting that the induction of H3K27M and p53 mutations at different stages of mouse development will likely distinct impacts on tumorigenesis. Induction of H3K27M also increase DIPG tumor formation induced by PDGFRA expression, but dramatically induce DIPG tumor formation and reduce mice survival when combined with both p53 depletion and PDGFRA overexpression (Figure 1B [47]. Together, these studies establish that the H3K27M mutation, while incapable of transforming neuro-precursor cells alone, accelerates the transformation and tumorigenesis induced by Trp53 loss and PDGFRA overexpression alone and in combination. In addition to its role in the initiation of DIPG formation, it has been shown recently that the H3K27M mutant is also required for maintenance of DIPG tumors. Deletion of the K27M mutant allele in DIPG tumor cells, while having minimal impact on the cell growth in vitro, prevents tumor formation when implanted in mice [48, 49].

H3K27me3 and PRC2 are involved in gene silencing. Therefore, one would expect that the significant reduction of H3K27 methylation in DIPG cells leads to a global increase in gene transcription. Instead, gene expression changes detected in mouse DIPG tumors occur predominantly at bivalent genes in the progenitor cells. The promoters of bivalent genes contain both an active mark (H3K4me3) and a repressive mark (H3K27me3) [47]. These bivalent genes, in general expressed at low levels, are poised for activation [50]. Therefore, the reduction of H3K27 methylation by H3.3K27M mutant proteins make these genes more susceptible for deregulation. These results are consistent with the idea that the PRC2 complex is re-distributed to poised enhancers in K27M mutant cells and that the H3K27M mutant proteins are low/absent at strong PRC2 sites with high levels of H3K27 methylation.

In addition to the global reduction of H3K27 methylation, early studies indicate that some genes gain/retain H3K27 methylation in DIPG cells [22, 26]. Moreover, GO pathway analysis indicate that genes with retain/gain of H3K27 methylation are enriched with cancer pathway. Therefore, it was proposed both the global reduction of H3K27 methylation and locus-specific gain/retention of H3K27 methylation are important for tumorigenesis of DIPG. Supporting this hypothesis, it has been shown that Ezh2 is required for the tumorigenesis of DIPG in mouse models [45, 46]. Mechanistically, it has been shown that the PRC2 and H3K27 methylation are important to silence tumor suppressor genes including p16 and WT1 in DIPG cells for the proliferation of tumor cells [42, 45, 46]. As another note, if DIPG is driven only by the reduction of H3K27 methylation, one would expect that Ezh2 inactive mutation can also promote DIPG. However, so far inactive mutations at Ezh2 have not been found in DIPG tumors despite the fact that inactive Ezh2 mutations have been identified in other tumor types including T-cell acute lymphocytic leukemia and myeloproliferative neoplasms [51, 52]. Together, these studies support the model that the global reduction of H3K27 methylation, while pronounced, is unlikely the only driver changes for DIPG tumors. Instead, both the global reduction and locus specific retention of H3K27 methylation are important for tumorigenesis of DIPG.

Potential therapeutic compounds for DIPG tumors

Currently, radiation therapy shows limited benefits for DIPG patients, and no chemotherapeutic agents show improved efficacy for DIPG tumors. Based on the recent discoveries on oncogenic mechanisms of DIPG tumors, various agents/tool compounds have been tested and found to be effective against DIPG patient-derived xenograft (PDX) lines in vitro and in mouse models (Table 1). In this section, we briefly discuss agents with known mechanistic insights and potential implications. Based on changes in PRC2 and H3K27 methylation mediated silencing pathway is disrupted in DIPG cells, various agents targeting different components of the pathway including inhibitors targeting H3K27 demethylase, JMJD3 [19] [53], H3K27 methyltransferase readers of H3K27ac [45], BMI1, a component of the PRC1 complex that works with PRC2 in gene silencing (Box 2) [54, 55] show anti-tumor activity in vitro and in PDX mouse models. Because inhibitors targeting Jmjd3 and Ezh2 likely have the opposite effect on H3K27 methylation in cells, these results indicate that a delicate balance of H3K27 methylation is established for the survival of DIPG tumor cells.

Table 1.

Experimental agents that show anti-tumor activity for DIPG.

Compounds	Targets	Potential mechanisms	Reference
GSK-J4	JMJD3	Increases H3K27 methylation	[19, 53]
Panobinostat	HDAC	Increases in histone acetylation and likely expression of ERV	[56]
GSK343, EPZ6438	EZH2	Induce the expression of tumor-suppressor genes including p16INK4A.	[45]
PTC209, PTC-596	BMI-1	1) Induce G1/S arrest. 2) Sensitize DIPG cells to radiotherapy.	[54, 55]
JQ-1 I-BET151	BRD4	1) Decrease transcription. 2) Induce cell cycle arrest. 3) Induce differentiation.	[32]
Thz1	CDK7	Disrupt RNAPII-dependent transcription.	[76]
MI-2	Lanosterol synthase	Disrupt cholesterol homeostasis.	[77, 78]
Anti-GD2 CAR T cells	Disialoganglioside GD2	GD2-targeting immunotherapy.	[57]
ONC201	DRD2	1) Deplete cancer stem cells. 2) Activate integrated stress response	[79]

Grasso et al reported panobinostat possess the anti-tumor activity in vitro and in animal models[56]. Recently, a combination of panobinostat and 5-azacytidine treatment of DIPG cells leads to increased expression of endogenous retrovirus (ERV), synergistic killing of DIPG cells in vitro and increased survival of mice transplanted with DIPG tumor cells[33]. Panobionstat is a pan inhibitor for histone deacetylases and 5-azacytitidine is an inhibitor targeting DNA methyltransferases. Therefore, a combination of panobionstat and 5-azacytitidine treatment likely compromises three major gene silencing mechanisms: DNA methylation, H3K9 methylation and H3K27 methylation in DIPG cells. Since DIPG cells with H3K27M mutation already show defects in H3K27 methylation mediated gene silencing, it would be interesting to determine whether unique agents for DIPG cells with the H3K27M mutation targeting these silencing mechanisms could be found.

In addition to small molecule inhibitors targeting potential “epigenetic regulators”, agents for tumor-immunotherapy also show promises in preclinical studies. For instance, H3K27M DIPG tumor cells were found to express high levels of cell surface marker, disialoganglioside GD2. GD2 Chimeric Antigen Receptor (CAR)-expressing T cells show potent anti-tumor activity in five independent PDX orthotopic mouse models [57] and thereby the promises of cancer immune therapy for DIPG.

In summary, it is exciting time for DIPG research as a variety of promising agents are found to be effective in killing DIPG tumors in preclinical studies. Future characterization of these agents in term of anti-tumor mechanisms and toxicity will provide additional guidance for the potential usage of these agents for the treatment of DIPG patients in the future. With increased understanding of DIPG tumorigenesis, we expect to discover additional drug targets/agents that will impact on the treatment of DIPG tumors in the future.

Concluding Remarks

In the last several years, we have witnessed an explosion of studies aimed at understanding the molecular mechanism of DIPG pathogenesis and discovering potential therapeutic modalities for this deadly disease. However, many questions remain to be addressed in order to understand this deadly disease (see outstanding questions). For instance, it remains unclear why is the H3K27M mutation is found almost exclusively in DIPG and midline gliomas, and how it occurs. It is possible that a certain neuro-precursor cells at this region of the brain during early development is more susceptible for cellular transformation by H3K27M mutant proteins. Clonal evolution model suggests that H3K27M mutation potentially arise first during DIPG tumorigenesis and exist in almost all subpopulation of DIPG tumors[58, 59]. Therefore, future studies aimed at identifications of cell of origin for the DIPG tumors and elucidation how these cells acquire H3K27M mutation will help answer this question. Second, it would be interesting to determine whether H3.1K27M and H3.3K27M mutant proteins impact tumorigenesis via distinct mechanisms. H3.1K27M, but not H3.3K27M DIPG tumors most likely contain additional mutations at ACVR1. H3.1K27M and H3.3K27M, like their wild type counterparts, are likely distributed on chromatin distinctly. Therefore, future studies aimed at understanding of the distinct impact of H3.1K27M and H3.3K27M on epigenomes will provide insights into distinct impact of H3.1K27M and H3.3K27M on epigenomes and the association of H3.1K27M mutation with the ACVR1 mutation for tumorigenesis of DIPG. Third, signal pathways that regulate the proliferation, cell death and migration of DIPG cells have not been well studied. Therefore, our efforts to identify signaling pathways that regulate these processes will likely help not only help understand oncogenic mechanisms, but also provide additional strategies for the treatment of DIPG tumors in the future. Finally, cancer immune therapy has shown promises for anti-DIPG activity. Therefore, an increased understanding of DIPG tumor microenvironment should be further to improve our ability to target DIPG cells for cancer immune therapy. With enthusiastic support from international communities and funding agency, significant insights into the tumorigenesis and treatment of DIPG tumors will be forth coming.

Outstanding questions.

• Why is H3K27M mutation found almost exclusively in in DIPG and midline gliomas?

• Do H3.1K27M and H3.3K27M mutations impact epigenome and promote tumorigenesis differently? Can we harness these differences for the treatment of H3.1K27M and H3.3K27M DIPG tumors with unique therapy?

• What are the interplays between signal pathways and H3K27M-induced epigenetic alteration for DIPG tumorigenesis?

• How H3K27M mutant proteins induce a conformation changes of the PRC2 complex in cells

Highlights.

• Over 60–70% DIPG tumors have lysine 27 to methionine mutations (K27M)

• Histone H3K27M mutations reprogramme epigenome with a global reduction of H3K27 methylation

• Sequestering PRC2 at poised enhancers by H3K27M mutant proteins represents a mechanism for the global reduction of H3K27 methylation

• H3K27M mutation accelerates tumorigenesis, especially for brainstem HGG and high grade DIPG