Oncohistones: exposing the nuances and vulnerabilities of epigenetic regulation

Summary:

Work over the last decade has uncovered a new layer of epigenetic dysregulation. It is now appreciated that missense somatic mutations in histones, the packaging agents of genomic DNA, are often associated with human pathologies, especially cancer. Although some of these ‘oncohistone’ mutations are thought to be key drivers of cancer, for the majority, the impact on disease onset and progression remains to be elucidated. Here we survey this rapidly expanding research field with particular emphasis on how histone mutants, even at low dosage, can corrupt chromatin states. This work is unveiling the remarkable intricacies of epigenetic control mechanisms. Throughout, we highlight how studies of oncohistones have leveraged, and in some cases fueled, advances in our ability to manipulate and interrogate chromatin at the molecular level.

Introduction

Studying the molecular basis of disease inevitably leads to a deeper understanding of fundamental biological processes. Examples include the seminal work on the tumor suppressor proteins p53 and BRCA, studies that revealed how cells sense genome integrity and respond to DNA damage (Williams and Schumacher, 2016; Roy et al., 2012), and on the Ras oncogenes, which showed how nature deploys binary molecular switches to regulate cell survival and proliferation (Simanshu et al., 2017). This theme extends to the field of epigenetics where, for example, studies of oncogenic fusions involving the MLL gene have yielded key insights into the mechanisms that regulate chromatin structure and consequently gene expression (Krivtsov et al., 2017). Indeed, the complexity of epigenetic regulation makes pulling on the ‘molecular thread’ of disease an especially effective means to uncovering key regulatory features of the epigenetic landscape and the biochemical mechanisms that underpin them.

Eukaryotic genomes are packaged into a nucleoprotein complex called chromatin. The repeating unit of this hetero-polymer is the nucleosome, which consists of ~146 bp of DNA wrapped around an octameric protein core comprised of two copies each of histones H2A, H2B, H3 and H4 (Luger et al., 1997). Chromatin satisfies two seemingly paradoxical requirements; it both stably condenses the genome such that its chemical integrity is maintained, yet also permits timely and locus-specific access to a myriad of nuclear factors required for DNA-templated transactions such as transcription. Deciphering how this balancing act is achieved with high fidelity down unique cell lineages is at the core of molecular epigenetics research. One key component of this elaborate choreography is the covalent modification of DNA and histones (Rothbart and Strahl, 2014) (Figure 1A). These ‘marks’ can alter the function of associated chromatin regions either directly, by changing the intrinsic structural properties of (poly)nucleosomes, or indirectly, by recruiting nuclear factors with ‘reader’ capabilities that help initiate the activation or repression of downstream processes. Consequently, the installation and removal of chromatin modifications, by dedicated ‘writer’ and ‘eraser’ enzymes, respectively, is tightly regulated and can involve direct post-translational modifications (PTMs) of the enzymes, changes in the levels of their metabolic cofactors, or the pre-existence of ‘marks’ on the substrates themselves (Yang and Li, 2020; Diehl and Muir, 2020; Torres and Fujimori, 2015). This final regulatory mechanism consists of both positive and negative biochemical crosstalk that occurs between chromatin modifications (Fischle et al., 2003). For example, many chromatin-associated enzymes employ a read-write/erase mechanism in which enzyme recognition of its product mark stimulates its own activity (Zhang et al., 2015). Collectively, these regulatory mechanisms couple dynamic changes in cellular metabolism and/or signal transduction to epigenetic regulation, providing a means by which even transient alterations in cellular state can have a lasting impact.

Figure 1.

Epigenetic regulation in health and disease. (A) Under normal physiological conditions, a combination of chromatin factors install (‘writers’), remove (‘erasers’), and recognize (‘readers’) DNA and histone PTMs, and remodelers function to regulate chromatin access. Such regulatory mechanisms ensure proper cellular responses to metabolic, environmental, and developmental cues. (B) Under pathophysiological conditions, including cancer, dysregulation of epigenetic machinery, via mutation or aberrant expression of chromatin factors, may contribute to disease development and/or progression. Adapted with permission from Nacev et al., 2019. Copyright (2019) Springer Nature.

Given the link between the epigenetic landscape and cellular phenotypes, mistakes made in establishing and maintaining chromatin states, as a consequence of abnormal inputs and outputs, are associated with a number of human pathologies (Fahrner and Bjornsson, 2014). In particular, the initiation and progression of many cancers has been linked to the dysregulation of a host of epigenetic regulators, including enzymes involved in DNA cytosine methylation and demethylation (Nishiyama and Nakanishi, 2021), writers of various histone marks and their eraser counterparts (Dawson and Kouzarides, 2012), transcription factors containing reader domains for histone marks (Mio et al., 2019), and ATP-dependent chromatin remodelers that regulate chromatin accessibility by mobilizing nucleosomes (Clapier, 2021) (Figure 1B). This final category includes the BRG1/BRM-associated (BAF) complexes that are mutated in over 20% of human cancers, a prevalence that suggests some form of epigenetic reprogramming is required during cell transformation (Kadoch et al., 2013). The strong correlation between epigenetic dysregulation and oncogenesis has fueled an ever-growing number of drug development programs (Helin and Dhanak, 2013), all of which seek to correct or exploit a perceived imbalance in epigenetic homeostasis. Inhibitors of histone deacetylases (HDACs) have been in use in clinical oncology for many years, and several other drug candidates targeting epigenetic factors show promise for treating a variety of disorders (Helin and Dhanak, 2013; Ho et al., 2020).

Historically, trans-acting factors such as histone writers and erasers have been the focus of research on epigenetic dysregulation in disease. By contrast, less attention has been paid to the substrates of these enzymes, namely the core histones, perhaps in part because they are encoded by multiple genes and have variant forms with specialized functions. For example, the replication-dependent histone variants, H3.1 and H3.2, are encoded by ten and three genes, respectively, and the replication-independent variant, H3.3, is encoded by two. Therefore, one might anticipate that somatic mutation of one histone allele might be inconsequential to disease etiology, given most histones in a mutant-expressing cell would be wild type. However, such reasoning was challenged about a decade ago following the discovery of somatic missense mutations in histone H3 genes that occur with high genetic penetrance in rare pediatric brain and bone cancers (Schwartzentruber et al., 2012; Wu et al., 2012; Behjati et al., 2013). These findings mobilized the epigenetics community to investigate if and how these mutants might alter cellular phenotypes (Deshmukh et al., 2021; Sahu and Lu, 2022). Thanks to the efforts of many laboratories, we now know that certain N-terminal histone H3 mutations function as cancer drivers (Hoffman et al., 2016; Nikbakht et al., 2016; Lu et al., 2016; Larson et al., 2019; Silveira et al., 2019; Harutyunyan et al., 2019; Khazaei et al., 2020). Remarkably, these so-called ‘oncohistones,’ despite their low abundance in chromatin, cause global reductions in key histone marks linked to transcriptional regulation (Lewis et al., 2013; Lu et al., 2016).

Although much still remains unknown regarding how the biochemical properties of oncohistones translate to disease-causing disruptions in the epigenetic landscape, their discovery challenged the assumption that dilution of a histone mutant within a wild-type ‘reservoir’ would mitigate any deleterious effects it might otherwise have. Clearly, there is something about the chromatin environment in which these mutants reside that can amplify their seemingly weak signal so as to promote profound dysregulation. Mechanistic studies of oncohistones, the numbers of which have increased significantly in recent years, offer a window into previously hidden aspects of epigenetic control. It is this idea—that oncohistones act as a sort of “stress test” for chromatin—that we will explore in this article.

How cancer ‘Met’ histones

In 2012, a series of groundbreaking studies reported the existence of recurrent hotspot mutations in pediatric high-grade gliomas (pHGGs) that map to the unstructured N-terminal tail of histone H3 (Khuong-Quang et al., 2012; Schwartzentruber et al., 2012; Wu et al., 2012). These studies were quickly followed by the discovery of equally prevalent H3 mutations, also in the histone H3 tail, in rare pediatric sarcomas such as chondroblastomas (Behjati et al., 2013). The high frequency of these mutations (e.g. H3K27M occurs in 80% of diffuse intrinsic pontine gliomas (DIPGs)) as well as their spatio-temporal specificity suggested a potential role in disease onset and/or progression. Shortly thereafter, several laboratories, including ours, discovered that H3K27M mutation in pHGGs results in a profound reduction in global levels of H3 with trimethylated lysine 27 (H3K27me3) (Venneti et al., 2013; Lewis et al., 2013; Bender et al., 2013; Chan et al., 2013), a histone mark associated with gene silencing and inherited through replication (Cao et al., 2002; Escobar et al., 2019; Hugues et al., 2020), with a concomitant, albeit more modest, increase in the levels of H3K27 acetylation (H3K27ac), a mark associated with active transcription. This was a remarkable finding given that the mutant represents only a small fraction (3–17%) of the total H3 pool in these cells (Lewis et al., 2013). Equally stunning was the discovery that ectopic expression of an H3.3K27M transgene in HEK293T cells also results in global reduction of H3K27me3 (Lewis et al., 2013; Bender et al., 2013; Chan et al., 2013), suggesting that the mutant itself is in some way responsible for reducing the levels of the mark on wild-type histones as well. Moreover, this gain of function effect was found to be restricted to methionine, and to a lesser extent isoleucine, as replacement of H3K27 with each of the other canonical amino acids had no effect on H3K27me3 levels (Lewis et al., 2013; Castel et al., 2015).

The H3K27me3 mark and the two lower methylation states at the same site (i.e. H3K27me1/2) are installed by the histone methyltransferase (HMT) polycomb repressive complex 2 (PRC2). The activity of this multi-subunit writer enzyme, which is itself frequently mutated in cancers (Comet et al., 2016), is regulated through the engagement of several auxiliary factors and by the presence of various pre-existing histone PTMs on the chromatin substrate (Holoch and Margueron, 2017). Perhaps most famously, the catalytic EZH2 subunit of the core complex is allosterically activated by binding of the H3K27me3 product to the regulatory EED subunit (Margueron et al., 2009). A similar mode of allosteric activation is associated with binding of the auxiliary subunit, JARID2 (Sanulli et al., 2015). This positive feedback is critical for so-called ‘spreading’ of H3K27me3 over extended regions of chromatin, a key component of gene silencing during development (Oksuz et al., 2018; Hugues et al., 2020). The sensitivity of PRC2 to various chromatin features and its exclusivity as the only known writer of H3K27 methyl marks motivated biochemical studies that revealed H3K27M-containing nucleosomes, and even H3K27M peptides, inhibit PRC2 enzymatic activity in a dose-dependent fashion (Lewis et al., 2013; Brown et al., 2014). Importantly, this inhibition is restricted to H3 proteins or peptides containing either methionine or isoleucine at position 27, a finding that mirrors those from heterologous expression studies in cells. Moreover, incorporation of norleucine (Nle), where the sulfur atom of methionine is replaced by a methylene, generates an even more potent peptide inhibitor of PRC2, suggesting enzyme inhibition requires a long hydrophobic sidechain. Photo-crosslinking experiments employing a diazirine-containing analog of methionine, photoMet, indicated that the inhibitor peptide binds exclusively to the EZH2 subunit (Lewis et al., 2013). Subsequent x-ray crystal structures of the PRC2 core complex bound to H3K27M peptides confirmed this finding and, in the case of the human enzyme, revealed that the methionine inserts into the substrate lysine binding pocket of the catalytic SET domain of EZH2 (Figure 2A). In other words, the H3K27M-containing peptide sequence acts as a competitive pseudosubstrate inhibitor of the enzyme (Jiao and Liu, 2015; Justin et al., 2016). In fact, tight binding H3K27M peptide inhibitors were key to obtaining diffraction quality crystals in these pioneering structural studies, likely due to the resulting stabilization of the SET domain of PRC2.

Figure 2.

PRC2 inhibition by H3K27M. (A) The crystal structure of human PRC2 in complex with a H3K27M peptide reveals how methionine insertion into the substrate-binding pocket of the EZH2 catalytic subunit provides a structural basis for pseudosubstrate inhibition by the histone mutant (PDB 5hyn). (B) PRC2 preferentially binds bivalent H3K27me3-H3K27M nucleosome units, leading to reduced enzyme activity and global changes in H3K27 marks.

The regulation of enzyme activity via cis- or trans-acting pseudosubstrate inhibitors is a well-known phenomenon especially common among protein kinases (Kobe and Kemp, 1999; Holub, 2020). Nevertheless, the discovery that a single amino acid substitution could convert a histone methyltransferase substrate into an orthosteric inhibitor was unprecedented. Naturally, this raised the question of whether the K-to-M ‘switch’ would have analogous effects in other contexts. In their initial report on the biochemical properties of H3K27M, Lewis et al. showed that heterologous expression of H3K9M and H3K36M in HEK293T cells results in global reduction of H3K9me2/3 and H3K36me2/3 levels, respectively (Lewis et al., 2013). Like H3K27M, these mutants directly inhibit the HMTs responsible for installing the marks, namely G9a and SUV39H1 in the case of H3K9me2/3, and Nsd2 and Setd2 for H3K36me2/3 (Lewis et al., 2013; Lu et al., 2016). Moreover, the pseudosubstrate HMT inhibitor paradigm is not restricted to the histone template. Recent studies have shown that global H3K27me2/3 levels are reduced in cells expressing a previously uncharacterized protein containing a C-terminal motif that closely mimics the H3K27M sequence (Pajtler et al., 2018; Hübner et al., 2019; Jain et al., 2019). This protein, which is overexpressed in posterior fossa group A (PFA) ependymomas, competitively inhibits the HMT activity of PRC2 and thus was named EZHIP (EZH2 inhibitory protein).

K-to-M mutation does not generate a pseudosubstrate inhibitor in every sequence context. For instance, heterologous transient expression of H3K4M, H3K79M, and H4K20M mutants in HEK293T cells does not alter global levels of their respective methyl marks (Lewis et al., 2013; Chan et al., 2013). Moreover, context seems to matter even in cases where the K-to-M ‘switch’ does have an effect. For example, H3K27me3 and H3K36me3 levels are not reduced upon expression of chimeric histones in which the corresponding mutant H3 tail is fused to the H4 core (note that this is not the case for chimeric H3K9M, expression of which does reduce H3K9me3) (Lewis et al., 2013). This implies that the surrounding chromatin framework is important, which is consistent with the fact that H3K27M nucleosomes are more potent PRC2 inhibitors than the corresponding peptides (Brown et al., 2014). Furthermore, location is not the only variable that influences inhibitor potency. K-to-M pseudosubstrate inhibitors bind more tightly to their corresponding HMT-SET domains when the S-adenosyl methionine (SAM) co-factor binding pocket is occupied (Justin et al., 2016; Jayaram et al., 2016; Jain et al., 2019). G9a, for example, binds H3K9M peptides ~16-fold more tightly in the presence of saturating concentrations of SAM (Jayaram et al., 2016). The situation with PRC2 is even more interesting since both H3K27M nucleosomes and EZHIP bind the enzyme with several-fold higher affinity in the presence of stimulatory peptides that engage EED (Stafford et al., 2018; Jain et al., 2019), suggesting, somewhat counterintuitively, that PRC2 must be allosterically activated for optimal inhibition. Building on these observations, biochemical studies with asymmetric designer chromatin templates revealed that PRC2 preferentially binds a bivalent inhibitor unit wherein H3K27M and H3K27me3 are located on different nucleosomes (Diehl et al., 2019) (Figure 2B). The strength of this bivalent inhibition decreases as a function of increasing internucleosomal distance, suggesting that PRC2 may bind tightest at boundary regions where H3K27me3 and H3K27M are in close proximity.

H3K27M-H3K27me3 interplay in cells: the plot thickens

Despite the considerable body of biochemical and structural work showing that H3K27M acts as a competitive inhibitor of PRC2 in vitro, it remains incompletely understood how such inhibition leads to epigenetic dysregulation in pHGGs. Collectively, the in vitro work suggests that PRC2 should be sequestered at H3K27M sites, preferentially so in contexts where the enzyme is allosterically activated (such as at H3K27M-H3K27me3 boundaries). However, several cell-based observations are difficult to reconcile with this mechanistic model. First, genome-wide mapping approaches employing chromatin immuno-precipitation with next generation sequencing (ChIP-seq), or the more recently introduced CUT&RUN method (Skene and Henikoff, 2017), do not reveal a strong co-localization of PRC2 and H3K27M, at least at steady state (Piunti et al., 2017; Harutyunyan et al., 2019; Sarthy et al., 2020). Second, multiple studies have shown that despite a global loss of H3K27me3 in H3K27M expressing cells, the mark is retained at many loci, especially at strong PRC2 recruitment sites where nucleation of H3K27me3 spreading occurs (e.g. densely unmethylated CpG islands) (Funato et al., 2014; Mohammad et al., 2017; Larson et al., 2019; Silveira et al., 2019; Harutyunyan et al., 2019; Sarthy et al., 2020). Levels of H3K27me3 in these residual domains are higher in H3.3K27M-expressing cells compared to those expressing H3.1K27M (Sarthy et al., 2020), possibly because the former has a more restricted genome distribution (Goldberg et al., 2010). It should also be noted that H3K27M DIPGs seem to acquire different secondary mutations depending on the clonal variant present; for example, H3.1K27M-DIPGs co-segregate with ACVR1 mutations, whereas H3.3K27M-DIPGs are often associated with TP53 pathway mutations (Wu et al., 2014; Mackay et al., 2017). These secondary mutations (e.g. PDGFRA, PDGFB) appear to synergize with K27M in tumor development (Funato et al., 2014; Mohammad et al., 2017). Further, in a recent study, Henikoff and co-workers suggest that dysregulation of developmental transcription factors may also contribute to the pathogenesis of K27M in a Drosophila model (Ahmad and Henikoff, 2021).

How might we reconcile these apparently incongruous biochemical and cellular observations? A recent series of papers from the Reinberg group offers a possible path out of this morass, while also illustrating how the study of oncohistones can reveal formerly unknown aspects of epigenetic regulation. The group used a doxycyclin-inducible H3.3K27M expression system to perform a ChIP-seq time-course in engineered HEK293 T-REx cells (Stafford et al., 2018). Surprisingly, given previous observations at steady state, these studies revealed that PRC2 is recruited to H3.3K27M-containing chromatin at early timepoints (up to 12 hours) following oncohistone expression. However, this co-localization is lost after 24 hours. Such transient engagement might be difficult to detect in ChIP-seq studies performed at steady state. The study also revealed that H3K27M histones acquire H3K36me2 at later timepoints. The H3K36me2 mark inhibits H3K27me3 deposition in cis (Yuan et al., 2011; Schmitges et al., 2011), potentially contributing, along with other histone marks such as acetylation (Piunti et al., 2017), to PRC2 eviction from H3K27M loci at later timepoints. Perhaps the most surprising finding was that PRC2 isolated from H3.3K27M-expressing cells has lower intrinsic HMT activity than that isolated from wild-type cells. Moreover, the same attenuation of activity is observed in recombinant PRC2 complexes pre-treated with H3K27M oligonucleosomes, implying that exposure to the oncohistone somehow irreversibly reduces enzyme activity. Clues as to the origins of this mysterious ‘memory’ effect come from subsequent work from this group in which they demonstrated that automethylation of key residues in the EZH2 subunit greatly enhance PRC2 HMT activity (Lee et al., 2019). Strikingly, this automethylation is reduced in HEK293 cells expressing H3K27M and in H3K27M-DIPG cells. This raises the intriguing possibility that EZH2 automethylation is somehow lowered as a consequence of the transient engagement of the enzyme with H3K27M chromatin, thereby accounting for the persistent attenuation of PRC2 activity in cells expressing the oncohistone. Whether this reduction is due to oncohistone inhibition of de novo automethylation or to loss of pre-existing automethylation while PRC2 is resident on H3K27M chromatin remains to be elucidated.

Collectively, the existing biochemical and genomic data support a model whereby exposure (most likely transient) of PRC2 to H3K27M-chromatin leads to a persistent reduction, but critically not elimination, of its HMT activity. Residual PRC2 activity may, in the end, be a key factor in directing epigenetic reprogramming in cells, since it sets a threshold either side of which different outcomes would be expected depending on the chromatin landscape. H3K27me3 would be retained in regions with strong PRC2 recruitment capabilities (and hence PRC2 density), whereas the mark would be lost around weaker nucleation sites. This threshold concept might also account for why complete inhibition of PRC2 using small molecules directed against EZH2 does not phenocopy oncohistone expression in cells (Mohammad et al., 2017; Harutyunyan et al., 2019). Indeed, pharmacological studies suggest that some level of PRC2 activity is necessary for pHGG tumor viability and that, by extension, inhibition of the enzyme may have therapeutic value (Mohammad et al., 2017). The model also predicts that alterations to the chromatin landscape itself could have therapeutic benefits, either by ‘detoxifying’ the H3K27M mutation (Brown et al., 2014) or by manipulating the installation or readout of other histone marks whose homeostasis is perturbed by the impact of the oncohistone on H3K27me3 levels (Yu et al., 2021). It is worth noting that efforts have been made to ameliorate the consequences of H3K27M through inhibition of various chromatin factors (e.g. HDACs) with some success, however the underlying mechanistic basis of these pharmacological effects remains somewhat unclear (Funato et al., 2014; Hashizume et al., 2014; Grasso et al., 2015; Piunti et al., 2017). In the end, any therapeutic strategy seeking to exploit epigenetic vulnerabilities created by H3K27M will have to navigate the crosstalk that occurs between histone marks.

H3K36M-H3K36me3 interplay in cells: one mark’s loss is another mark’s gain

Some 95% of chondroblastomas have a heterozygous mutation in the H3F3B gene that leads to expression of H3.3K36M (Behjati et al., 2013). This is the only recurrent mutation in these sarcomas, making it a prime cancer driver candidate. Furthermore, H3K36M mutations have also been observed in certain soft tissue sarcomas and head and neck cancers (Shah et al., 2014; Lu et al., 2016; Papillon-Cavanagh, 2017). Like H3K27, lysine 36 of histone H3, can be mono-, di-, or tri-methylated, yet unlike H3K27, different HMTs are associated with each methylation state. H3K36me1/2 can be installed by multiple SET-domain containing enzymes including Nsd1, Nsd2 and Setd2, whereas Setd2 is the only enzyme known to convert H3K36me2 to H3K36me3 (Wagner and Carpenter, 2012). Whereas H3K27me3 is associated with facultative heterochromatin, H3K36me2/3 marks are enriched within actively transcribed gene bodies where they prevent cryptic transcription initiation by recruiting the de novo DNA methyltransferases, DNMT3A and DNMT3B (Carrozza et al., 2005; Neri et al., 2017; Weinberg et al., 2019). The marks also have roles in splicing and DNA repair (Wagner and Carpenter, 2012; Sun et al., 2020).

Chondroblastoma tumor samples expressing H3.3K36M exhibit a global reduction in H3K36me2/3 levels, an effect also observed in various cell lines ectopically expressing the mutant, including chondrocytes and mesenchymal progenitor cells (MPCs) (Fang et al., 2016; Lu et al., 2016). Expression of H3.3K36M in MPCs blocks their ability to differentiate down osteocytic and adipocytic lineages (Lu et al., 2016). Moreover, immunodeficient mice injected with H3.3K36M-MPCs develop sarcomas, providing strong evidence that this mutant acts as a cancer driver (Lu et al., 2016). There is considerable evidence suggesting H3.3K36M follows the pseudosubstrate inhibitor paradigm of other K-to-M mutants. As mentioned earlier, in vitro biochemical studies indicate that H3.3K36M directly inhibits the HMT activity of Nsd2 and Setd2 (Lu et al., 2016; Fang et al., 2016). Structural studies show that the methionine fits snugly into the substrate binding channel of the Setd2 catalytic domain (Yang et al., 2016; Zhang et al., 2017) (Figure 3A). Moreover, Nsd2 and Setd2 co-immunoprecipitate with H3K36M nucleosomes isolated from MPC cells (Lu et al., 2016). These findings have led to a model in which H3K36M binds and sequesters the H3K36me2/3 HMTs in cells leading to the observed reduction in the marks (Lu et al., 2016).

Figure 3.

H3K36 methyltransferase inhibition by H3K36M and H3G34 mutations. (A) The crystal structure of Setd2 in complex with a H3.3K36M peptide reveals insertion of methionine into the substrate binding channel of the enzyme. Mutation of G34 to bulkier amino acids would be expected to lead to steric clashes with the protein (PDB 4h12). (B) Both H3K36M and H3G34R/V/W/L mutants inhibit H3K36 HMTs, but the former results in global decreases in H3K36 methyl marks whereas the latter exerts its effects on mutant histone K36 methylation only.

A global reduction in H3K36me2/3 levels would be expected to impact gene regulation. Indeed, expression of H3.3K36M in chondrocytes and MPCs leads to altered transcriptional profiles (Lu et al., 2016; Fang et al., 2016), which, in the case of latter, includes reduced expression of several master regulators of adipogenesis and osteogenesis as well as increased expression of transcription factors associated with maintaining the mesenchymal state (Lu et al., 2016). Notably, these transcriptional changes are mirrored by depletion of the H3K36 HMTs. Given the diverse roles of H3K36me2/3 in genome regulation, the mechanisms driving these transcriptional changes are likely complex and multifaceted. As previously mentioned, H3K36me2/3 inhibits installation of H3K27me3 on the same histone tail (Yuan et al., 2011; Schmitges et al., 2011). Thus, in H3.3K36M-MPCs, a global reduction in H3K36me2/3 levels results in increased H3K27me3 levels due to a loss of this cis inhibition (Lu et al., 2016). Further analysis of this crosstalk using ChIP-seq revealed that whereas H3K36me2/3 levels decrease genome-wide, the H3K27me3 mark increases primarily at intergenic regions previously marked by H3K36me2/3. This ‘PTM seesaw’ leads to a redistribution of polycomb repressive complex 1 (PRC1), which binds H3K27me3 and suppresses gene expression (Schuettengruber et al., 2017), from genic regions to newly established H3K27me3 intergenic loci (Figure 3B). This results in the derepression of PRC1-associated genes, many of which are linked to mesenchymal cell renewal. Whether this is the only mechanism driving the developmental block associated with H3.3K36M remains unclear, particularly since human chondrocytes expressing the oncohistone do not exbibit an obvious increase in global H3K27me3 levels (Fang et al., 2016). Nevertheless, the work in MPCs does reveal how histone marks in intergenic regions can impact the chromatin landscape elsewhere, in this case by altering the distribution of reader proteins.

Research on H3K36M has some bearing on another set of oncohistone mutants, namely those centered on Gly34 of histone H3. Unlike the K-to-M mutants discussed above, which occur on both replication-dependent and replication-independent H3 isoforms, the G34 mutants are only found on the H3.3 variant (Behjati et al., 2013). This suggests that localization of the mutants to actively transcribed genes and/or constitutive heterochromatin, regions where H3.3 is deposited by dedicated chaperones (Goldberg et al., 2010; Szenker et al., 2011), might be integral to pathogenesis. The H3.3G34 mutants differ from H3K27M and H3K36M in several other respects as well. First, Gly-34 is not a site of post-translational modification. Second, different H3G34 mutations are associated with different tumors; H3.3G34R/V mutations are found in pHGGs of the cerebral cortex, whereas H3.3G34R/W/L mutations are associated with giant cell tumors of the bone (GCTBs) and osteosarcomas (Behjati et al., 2013; Koelsche et al., 2017). Third, H3.3G34 mutants do not act in a dominant negative fashion to alter global levels of histone PTMs. Rather, their effect appears to be local, resulting in a reduction of H3.3K36 methylation levels on the same histone tail (Lewis et al., 2013; Zhang et al., 2017; Shi et al., 2018) (Figure 3B). Structural studies of Setd2 bound to H3K36M peptides indicate that mutation of G34 to bulkier amino acids would lead to steric clashes with the protein, presumably explaining the in cis reduction in H3K36 methylation (Yang et al., 2016; Zhang et al., 2017) (Figure 3A). The added steric bulk of G34 mutations may also impede the binding of other factors to this region of the histone tail. Indeed, this steric occlusion mechanism has been invoked in the case of the transcriptional repressor ZMYD11 and the histone demethylases KDM2A and KDM4 (Wen et al., 2014; Cheng et al., 2014; Voon et al., 2018).

Compared to the K-to-M oncohistones discussed above, much less is known about the effects of H3.3G34 mutants on epigenetic programs. pHGGs carrying H3.3G34 mutations have telomeric defects, perhaps due to reduced DNA methylation within these regions (Sturm et al., 2012). Recent studies employing engineered neuronal and mesenchymal progenitor cells have linked H3.3G34 mutation to several abnormalities, including altered splicing (Lutsik et al., 2020; Funato et al., 2021) and a redistribution of H3K27me3 towards genic regions enriched in H3.3 (Khazaei et al., 2020; Chen et al., 2020). The latter is thought to result from disruption of the normal PRC2 cis inhibition mechanism on the mutant histones, leading to a redistribution of the enzyme and eventually, through a currently ill-defined series of epigenetic remodeling events, a blockade in progenitor cell differentiation.

The expanding oncohistone landscape

The oncohistone mutations presented thus far were first identified largely on an individual basis in clinical settings due to their high disease penetrance. In the years following, the same mutations were noted in other cancers (e.g. H3K36M in head and neck cancers, (Papillon-Cavanagh et al., 2017)) and additional cancer-associated histone mutations were identified as well (e.g. histone H1 mutations in diffuse large B-cell lymphomas and H2A and H2B mutations in carcinosarcomas, (Lohr et al., 2012; Zhao et al., 2016)). In 2019, efforts were made by two groups to globally catalogue and characterize the histone missense mutational landscape across multiple cancer types (Nacev et al., 2019; Bennett et al., 2019). Somatic histone missense mutations were identified using publicly available data from cBioPortal and/or previously unreported institutional cancer genomics data (MSK-IMPACT) and documented to occur in ~4% of tumor samples—a number that probably represents an underestimate given that the MSK-IMPACT database contained less than one-third of the total histone genes. This mutational prevalence rivals that of cancer-associated genes within the same cohort, including BRCA2, TET2, SMAD4, and NOTCH1. As mentioned earlier, each of the core histones are encoded by multiple genes, perhaps contributing to previous failures to recognize their mutational abundance in rank-ordered lists of genes from tumor sequencing studies and a corresponding lack of emphasis on their study despite their high mutational rates. These bioinformatics analyses not only re-identified the founding set of oncohistones (i.e. H3K27M, K36M, G34R/V/W/L) but also revealed mutations across all four core histones in both the tails and globular domains (Nacev et al., 2019; Bennett et al., 2019). Critically, these new mutations differ from canonical oncohistones in three key respects: (1) native histone residues are not always mutated to one specific other amino acid (e.g. contra K36M); (2) these mutations are broadly distributed over many cancer types and are not as highly penetrant; and (3) many do not occur at or near sites of post-translational modification, suggesting that they might confer their effects through distinct mechanisms.

Big trouble at the core

Among the most unexpected findings from these cataloguing efforts was that four of the five most commonly mutated residues are found not in the histone tails but rather in the histone globular domains that form the core of the nucleosome. The most well-studied of these to-date are mutations at H2BE76, the third most commonly mutated site overall and first-most on H2B (Nacev et al., 2019). These mutations co-occur with common oncogenes such as RAS, PI3K, KDM6A, KMT2C, and TP53 (Bennett et al., 2019) and are especially prevalent in bladder and cervical cancers (Nacev et al., 2019). H2BE76 is located at the dimer-tetramer interface where it engages in hydrogen-bonding with H4R92 and D68, which are also highly mutated residues on H4 (Nacev et al., 2019) (Figure 4A). The most common amino acid mutations at H2BE76, namely E76K and E76Q, are predicted to disrupt these interactions. In the case of the E-to-K mutant, the lysine side chain causes electrostatic repulsions with the side chain of H4R92, ultimately changing the configuration of the H4 α3-helix (Arimura et al., 2018). Such structural perturbations would be expected to destabilize the nucleosome, a hypothesis borne out by the fact that H2BE76K histones fail to form octamers when combined with their complementary histones (Arimura et al., 2018; Bennett et al., 2019; Bagert et al., 2021). Consistent with this, H2BE76K- and, to a lesser extent, H2BE76Q-containing nucleosomes show decreased thermal stability and increased Nap1-mediated dimer exchange in vitro in both symmetric and asymmetric contexts (Arimura et al., 2018; Bagert et al., 2021) (Figure 4B). Analogous destabilizing effects were observed in cellular settings. For example, yeast expressing H2BE76K/Q exhibit temperature sensitivity as well as increased chromatin sensitivity to MNase (Bennett et al., 2019). FRAP experiments in mammalian cells revealed that the recovery kinetics of H2BE76K are faster than that of wild type, indicative of more rapid exchange of the mutant in and out of chromatin (Arimura et al., 2018; Bennett et al., 2019). Surprisingly, such destabilizing mutants appear to endow cells with common pro-proliferative phenotypes. MCF10A cells stably expressing H2BE76K show enhanced growth, and when the mutant is co-expressed with a PI3K mutant with which it co-occurs in breast cancer, the cells exhibit increased colony formation in soft agar (Bennett et al., 2019). Transient transfection of NIH3T3 cells with H2BE76K leads to a similar cell phenotype marked by reduced contact inhibition (Arimura et al., 2018). MPCs stably expressing H2BE76K/Q, or another dimer-tetramer interface mutant H2BE71K/Q, upregulate pathways implicated in tumorigenesis (e.g. PI3K-Akt, Wnt, Rap1) and downregulate those involved in cell adhesion (Bagert et al., 2021). Moreover, charge-swap (E-to-K), but not charge-neutralizing (E-to-Q), mutations inhibit cell differentiation to chondrocytes or adipocytes, consistent with the existing biochemical data demonstrating the former histone mutant to be more destabilizing than the latter (Bagert et al., 2021).

Figure 4.

H2BE76 mutation destabilizes nucleosomes. (A) H2BE76 is located at the dimer-tetramer interface where it engages in hydrogen-bonding with H4R92 and D68. Histone H2B is depicted in orange and histone H4 in green on the nucleosome structure (PDB 1kx5). (B) H2BE76K mutation decreases nucleosome thermal stability and enhances Nap1-mediated dimer exchange.

A mechanistic understanding of how destabilization of nucleosomes might manifest as upregulation of cancer-associated pathways and undifferentiated cell phenotypes has yet to be fully elucidated. However, correlative sequencing studies have begun to shed light on the subject. ATAC-seq analysis of MCF10A cells stably expressing physiologically relevant levels of H2BE76K revealed that >20% of new peaks resulting from histone mutant expression are found in what are normally heterochromatic regions. Increased accessibility was observed in the promoter regions of >3,000 genes which were also increasingly expressed, suggesting that destabilizing histone mutant incorporation might lead to enhanced genomic accessibility and increased transcription (Bennett et al., 2019). Yet when H2BE76K was knocked in to one copy of H2B in the MDA-MB-231 breast cancer cell line, no global chromatin accessibility changes were observed (Kang et al., 2021). Scrutinization of one specific upregulated target, ADAM19, in both wild-type and mutant cell lines revealed similar RNA Pol II occupancy but faster elongation in the case of the mutant cells at this site (Kang et al., 2021). This study suggests that mutant incorporation at specific loci prompts increased transcription of target genes not by altering nucleosome occupancy but rather by facilitating RNA Pol II transcription, since these destabilizing mutant-containing nucleosomes pose an inherently lower barrier to transcription than wild-type ones. Precisely how preferential localization of H2BE76K mutant histones to specific regions of chromatin might occur and whether that is a consequence of targeted deposition or a steady-state phenomenon brought about indirectly remains unclear. However, what is evident is that H2BE76K mutations can alter cell fates, predisposing them to cancer-promoting phenotypes (Arimura et al., 2018; Bennett et al., 2019; Bagert et al., 2021; Kang et al., 2021). Interestingly, unlike H3K27M and K36M mutant cell lines, no global changes in PTMs have been documented in H2BE76K-expressing cell lines to date, although it should be noted that the increased nucleosomal dynamics associated with the mutation might be expected to alter access to modifiable residues that would otherwise be sterically occluded. Collectively, these studies put forth dimer-tetramer interface mutants as a new class of oncohistone with mechanistic underpinnings apparently distinct from its forerunners. Other dimer-tetramer interface mutants abundant in cancers, including H2BD68, F70, H4D68, and K91, have also been shown to destabilize nucleosomes in vitro (Bagert et al., 2021). Whether these also contribute to altered cellular phenotypes has yet to be determined.

Another mutational hotspot identified from these global sequencing studies is the nucleosome acidic patch. This region, consisting of eight acidic residues in H2A and H2B (Luger et al., 1997), is a negatively charged surface present on both “faces” of the nucleosome (Figure 5A). The acidic patch has been shown to be important for the activity of numerous chromatin factors, which typically interact with the region via a positively charged “arginine anchor” motif (McGinty and Tan, 2021). Thus, the acidic patch is important for a variety of nuclear processes, including transcriptional silencing, nucleosome remodeling, cell division, chromatin condensation and folding, and DNA damage repair (McGinty and Tan, 2021). Remarkably, cancer-associated acidic patch mutants can either inhibit or activate ISWI-mediated nucleosome remodeling (Dann et al., 2017; Dao et al., 2020; Bagert et al., 2021). Specifically, mutations within the arginine anchor pocket (e.g. H2AE61D, D90N, E92K) decrease SNF2h/ACF complex-mediated nucleosome sliding, whereas acidic patch mutant residues more distal to this region (e.g. H2AE56K/Q and H2BE113K) increase remodeler activity (Dao et al., 2020; Bagert et al., 2021). Acidic patch mutations can also affect the activities of other classes of remodelers, including the mammalian SWI/SNF (or BAF) complexes. For example, mutation of acidic patch residues H2AE64, E91, and H2BE113 abrogates binding of the SMARCB1 subunit of BAF (Valencia et al., 2019), which directly engages the acidic patch via a C-terminal helical domain containing basic residues (He et al., 2020; Mashtalir et al., 2020). Indeed, acidic patch mutants decrease the remodeling rates of all three final form mSWI/SNF complexes (i.e. cBAF, PBAF, and ncBAF) (Mashtalir et al., 2021). Such mutants also engender altered cellular phenotypes such as growth defects in yeast and altered transcriptional profiles in mouse MPCs (Bagert et al., 2021). Mutation of H2BE113, the fifth most commonly mutated histone residue (Nacev et al., 2019), to Lys even inhibits MPC differentiation much akin to the well-recognized H3K36M oncohistone, suggesting that acidic patch mutants have the capacity to alter cell fates and potentially contribute to cancer (Bagert et al., 2021).

Figure 5.

Acidic patch mutations alter chromatin remodeler activities. (A) Acidic patch residues (red) on histones H2A (yellow) and H2B (orange) form a negatively charged surface on each face of the nucleosome (PDB 1kx5). (B) Acidic patch mutations alter ATP-dependent chromatin remodeler activities and affect nucleosome positioning.

Determining how disruption of chromatin remodeler activity contributes to dysregulation of gene expression represents a formidable challenge, but one that is beginning to reveal hitherto underappreciated aspects of chromatin regulation. Specifically, studies of the impact of acidic patch mutants on chromatin remodeler activity have led investigators to consider the role of molecular symmetry in chromatin structure. The polygenic nature of histone genes means that random incorporation of a low-abundance oncohistone mutant will generate heterotypic nucleosomes (i.e. containing one wild-type and one mutant copy), by definition breaking the intrinsic two-fold pseudo symmetry of the particle. This desymmetrization is especially interesting in the case of acidic patch mutants, since they afford nucleosomes with two distinct faces—what we have referred to as ‘Janus’ bioparticles (Dao et al., 2020; Mitchener and Muir, 2021). Certain chromatin remodelers engage both faces of the nucleosome (e.g. ISWI and mSWI/SNF complexes) which, in the case of the ‘Janus’ species, would each be unique, raising the possibility of differential interactions and consequently altered remodeling. Using recently developed chemical biology approaches to synthesize asymmetric nucleosomes (reviewed elsewhere, (Mitchener and Muir, 2021)), we and others determined that ISWI slides asymmetric nucleosomes with inhibitory acidic patch mutations unidirectionally to give longer linker DNA proximal to the mutant acidic patch face (Levendosky and Bowman, 2019; Dao et al., 2020) (Figure 5B). Conversely, ISWI moves asymmetric nucleosomes with stimulatory acidic patch mutations unidirectionally to give longer linker DNA proximal to the wild-type face (Dao et al., 2020). In a dinucleosome context, this functional imbalance caused by mutant incorporation leads to aberrant nucleosome spacing (Dao et al., 2020). Interestingly, acidic patch mutants can also reprogram cBAF remodeling activity but the trend is opposite that of ISWI—that is, cBAF translocates inhibitory asymmetric acidic patch mutant nucleosomes (e.g. H2AE92K) to generate longer linker DNA proximal to the wild-type face (Dao et al., 2022) (Figure 5B). Collectively, these studies demonstrate that cancer-associated histone acidic patch mutations can alter chromatin remodeling activities and affect nucleosome positioning in vitro. Further studies are needed to determine whether such effects translate to altered chromatin features, such as nucleosome occupancy and chromatin accessibility, in cells. More generally, this work provides yet another example of how the study of oncohistones can illuminate some of the more nuanced aspects of chromatin regulation, in this case the role of nucleosome (a)symmetry.

A final category of abundant cancer-associated histone core mutations consists of those that occur at sites of DNA-histone contact. Such mutations are found in basic residues (e.g. H2AR29, K74, K75, R77, and H4R45) in close proximity to the negatively charged DNA encircling the histone octamer (Figure 6A). Perhaps unsurprisingly, given the conserved mechanism of chromatin remodelers in harnessing the energy of ATP hydrolysis to break electrostatic interactions between histone amino acid side chains and DNA to promote nucleosomal translocation (Clapier et al., 2017), neutralizing mutations at these sites enhance nucleosome sliding by ISWI remodelers (Dann et al., 2017; Bagert et al., 2021). DNA-histone contact mutants have also been shown to have consequences in cells, as these mutations can cause growth defects in yeast (Maruyama et al., 2006; Bagert et al., 2021). One such mutant, H2BG53D, is located close to the DNA entry/exit site. Single-molecule nucleosome pulling assays reveal that although the unwrapping rate of DNA from wild-type and H2BG53D nucleosomes is similar, the wrapping rate of H2BG53D is four times slower, suggesting there is decreased interaction between the histone dimer and DNA (Wan et al. [1], 2020) (Figure 6B). Consistent with this finding and the notion that nucleosomes serve as inherent barriers to transcription, chromatin templates containing H2BG53D nucleosomes show increased RNA Pol II passage compared to their wild-type counterparts in in vitro transcription elongation assays (Wan et al. [1], 2020). When H2BG53D is knocked in at an endogenous locus in the S2VP10 pancreatic cell line, the resultant cells exhibit a strong correlation between mutant localization and increased gene expression (Wan et al. [1], 2020; Wan et al. [2], 2020). Moreover, mutant cell lines show enhanced gap closure and transwell migration, at least in part due to the upregulation of genes involved in cell migration (Wan et al. [1], 2020; Wan et al. [2], 2020). Whether other DNA-histone contact mutations confer their effects analogously by promoting transcriptional elongation, by altering chromatin states via changed interactions with chromatin remodelers (more akin to acidic patch mutations), or by another mechanism altogether remains to be seen.

Figure 6.

DNA-histone contact mutations enhance chromatin remodeling and decrease histone-DNA interactions. (A) Location of DNA-histone contact mutations (red) on the nucleosome structure. Histone H2A is depicted in yellow, histone H2B in orange, and histone H4 in green (PDB 1kx5). (B) H2BG53D mutation promotes ISWI-mediated chromatin remodeling and decreases DNA-histone interactions.

There are other ‘tails’ to tell

Although many of the highly abundant recently catalogued histone mutations occur in the histone globular domains, some previously unstudied histone tail mutations are also quite prevalent in cancers. In addition to the aforementioned canonical oncohistones, another tail residue among the top ten most mutated sites is H3R26 (Nacev et al., 2019). Due to their proximity to H3K27, H3R26 mutations might be anticipated to affect chromatin writers, readers, and/or erasers operating in the vicinity, and indeed, the PRC2 inhibitory activity of H3K27M peptides is highly sensitive to even minor modifications at H3R26 (Brown et al., 2014). N-terminal mutations in H3, including R26 but also R2, R8, and K18, were found in tumors with very low mutational backgrounds (TMB ≤2), suggesting potential roles for these mutants in facilitating cancer development (Nacev et al., 2019). H3K4 mutations, unlike the former, are predominantly of a specific amino acid identity—namely K-to-M/I (Nacev et al., 2019). Although ectopic expression of H3.3K4M in mouse preadipocytes does prevent their differentiation (Jang et al., 2019), unlike canonical K-to-M oncohistone mutations, H3K4M mutants do not appear to differentially engage the H3K4-specific methyltransferases MLL3 and MLL4 (Burton et al., 2020). However, in situ chromatin interactomics studies using a non-oxidizable Met mimetic, H3K4Nle, did reveal differential binding partners for the mutant, including increased association with the E3 ligase CUL4B and nuclear factor THOCH as well as decreased association with DNM3B, PHF13, and JADE1 (Burton et al., 2020). Also prevalent are mutations on the far N-termini of H2A and H4, especially those at H4S1, R3, G4, and H2AS1 and R3 (Nacev et al., 2019). Peptides bearing mutations at these residues show reduced acetylation by the H2A/H4-specific N-acetyltransferase NatD (Ho and Huang, 2022). Activity decreases result from decreased K_m values of the mutant histone peptides (Ho and Huang, 2022), suggesting compromised binding consistent with the NatD crystal structure that reveals a narrow binding channel likely to be blocked by substrate N-terminal mutation (Magin et al., 2015).

Cancer-associated histone mutations have also been identified in linker histone H1, which binds nucleosomes and facilitates chromatin compaction (Fyodorov et al., 2018). H1 mutations occur in 30% of follicular lymphomas, 30–40% of diffuse large B cell lymphomas, and 50% of Hodgkin lymphomas (Okosun et al., 2014; Li et al., 2014; Reichel et al., 2015). The vast majority of these (97%) encode missense mutations that affect the histone H1 globular or C-terminal domains. The former mutants exhibit nucleosome aggregation and bind with lower affinities to mononucleosomes, whereas the latter localize to chromatin but impair chromatin compaction, suggesting both types are loss-of-function mutations but with distinct mechanisms (Yusufova et al., 2021). Knockout mice lacking genes H1C and H1E, the most abundant and often concurrently mutated genes encoding H1, exhibit enlarged and more abundant germinal B cells (Yusufova et al., 2021). These cells display extensive decompaction of chromatin domains concomitant with a gain of accessibility in these regions, ultimately resulting in transcriptional activation of stem cell genes and reversed silencing of developmental PRC2 targets. Lymph nodes of H1c^−/−H1e^−/− mice are infiltrated with immunoblastic cells marked by large nuclei, open chromatin, and increased H3K36me2 levels. Moreover, knockout mice exhibit extensive invasion of the liver and lungs by neoplastic cells, an expanded monoclonal B cell population, and shorter survival times than their wild-type counterparts. Collectively, these data suggest that cancer-associated H1 mutations render the linker histones biochemically sub- or non-functional, resulting in three-dimensional decompaction of chromatin and enhanced self-renewal of lymphomas in vivo.

Summary

It has been ten years since the seminal reports of histone mutations occurring with high genetic penetrance in cancers. Over that time, the field’s understanding of how these low-abundance mutations confer a dominant effect has continued to evolve (Figure 7). We now know that H3K27M acts as a pseudosubstrate inhibitor of PRC2, irreversibly decreasing, but not completely eliminating, its activity, resulting in global PTM changes. H3K36M likewise inhibits its cognate methyltransfereases, also altering global PTMs and ultimately leading to a redistribution of the H3K27me3 reader PRC1. Alternatively, H3G34 mutations exert their effects in cis, inhibiting the activities of methyltransferases acting on adjacent PTM sites. In the past three years, there has been an explosion of newly identified histone mutants present in cancers, the biochemical and cellular consequences of which have just begun to be elucidated. Unlike canonical oncohistones, many of these histone mutations occur in the globular domains and may not alter global PTMs. Rather, biochemical studies have revealed various consequences of these mutations, including destabilization of nucleosomes, alteration of chromatin remodeler activities, disruption of DNA-histone interactions, and perturbation of chromatin compaction. Although likely mechanistically distinct from their forerunners, many still disrupt epigenetic signaling and transcription despite representing a small fraction of the total histone pool. Moreover, some of these histone mutants are associated not only with cancer but also with developmental disorders (Tessadori et al., 2017; Tessadori et al., 2020; Bryant et al., 2020). Thus, understanding how these new classes of oncohistones disrupt the epigenetic landscape is expected to provide insight into diseases beyond cancer as well. Doubtless many questions remain, including if such mutants affect histone PTMs, how and where they are deposited in chromatin, and the means by which their incorporation leads to specific gene expression changes. If the past is any indicator, the next decade will be fertile ground for oncohistone research and, consequently, for improved understanding of the epigenetic basis of disease. Along the way, we anticipate that by continuing to pull on ‘molecular thread’ afforded by oncohistones, new aspects of epigenetic regulation will be laid bare.

Figure 7.

Oncohistones in epigenetic dysregulation. Studies so far have revealed three major modes of action for oncohistones depending on the position and type of mutation: (i) mutations can reduce the intrinsic stability of nucleosomes; (ii) mutations can alter the activity of ATP-dependent chromatin remodelers, and; (iii) mutations can create pseudosubstrate inhibitors of histone PTM writers.

In this review, Mitchener and Muir discuss our evolving understanding of how cancer-associated histone mutations contribute to disease. They provide historical context for these discoveries, highlighting how the study of ‘oncohistones’ has provided mechanistic insights into epigenetic regulation.