Leveraging dominant-negative histone H3 K-to-M mutations to study chromatin during differentiation and development

Differentiation generates astonishingly diverse cell types, largely without altering DNA sequence, suggesting that exquisite regulation exists to interpret the genome in different ways. Epigenetic regulation (a mechanism that influences cell phenotype without altering DNA sequence) is thought to play a crucial role in guiding and reinforcing cell fate change (Greer and Shi, 2012), but our understanding of the factors involved remains incomplete. Histone proteins are prime candidates to mediate this epigenetic role for several reasons: histones can be extensively modified through the addition or removal of chemical groups, providing a versatile platform to display regulatory signals (Phanstiel et al., 2008). Furthermore, many histone modifications have been linked to transcriptional activation or repression (Greer and Shi, 2012), providing a mechanism by which histones can influence complex cell fate decisions (Zhou et al., 2011). Finally, histone modifications can be added or removed quickly, offering a dynamic system to fine-tune gene expression on a timescale that is compatible with rapid developmental changes (Brumbaugh et al., 2011; Apostolou and Hochedlinger, 2013; Perino and Veenstra, 2016).

Histone modifications influence transcription programs by altering chromatin structure and recruiting regulatory proteins to specific genes (Strahl and Allis, 2000; Mitchener and Muir, 2022). Consequently, knowing the precise genomic location of a particular histone modification provides important clues for establishing its function. Many approaches exist for profiling histone modifications, both at specific loci and genome-wide (Gilmour and Lis, 1984; Brind'Amour et al., 2015; Kaya-Okur et al., 2019). Comparing histone modification profiles to transcriptomic data provides a correlative link between a given mark and gene expression; however, some enzymes that modify histones directly interact with transcriptional machinery (Kizer et al., 2005). Thus, it remains an open question whether changes to histone modifications are a cause or a consequence of gene expression changes. In Drosophila, where replication-dependent histones are clustered, it is possible to knockout endogenous histone H3 genes and complement the system with a mutant version (Günesdogan et al., 2010; McKay et al., 2015). This approach has provided important insight into the role of histone modifications during development in flies (Pengelly et al., 2013; Penke et al., 2016); however, directly perturbing a histone modification to study its function is challenging in mammals, in part because histones are encoded many times in the genome (Sankar et al., 2022). As a result, the functional significance of particular histone marks has largely been inferred from gain- and loss-of-function studies for histone-modifying enzymes. A general limitation of these studies is that histone modifications are typically regulated by multiple enzymes, which are difficult to perturb simultaneously and may compensate for one another (Inagawa et al., 2013), as summarized in Fig. 1 in the context of histone H3 methylation. Moreover, these proteins often have non-histone substrates or act to recruit additional regulatory factors, raising the possibility that phenotypes observed in these studies are unrelated to histone marks (Miller et al., 2010; Xu et al., 2012; Warrier et al., 2022). Together, these challenges illustrate the need for tractable tools to directly manipulate specific histone modifications.

In 2012, recurrent mutations were identified on histone H3 in pediatric gliomas (Schwartzentruber et al., 2012; Wu et al., 2012). Common among these mutations were heterozygous, gain-of-function lysine-to-methionine (K-to-M) substitutions on residue 27 of H3 (H3K27M). Further research demonstrated that expression of H3K27M led to global suppression of H3K27 methylation on the remaining endogenous copies of H3 (Bender et al., 2013; Chan et al., 2013a,b; Lewis et al., 2013). In cancer, H3K27M mutations are primarily found on histone variant H3.3, a replication-independent histone, though K-to-M mutations on replication-dependent variants H3.1 and H3.2 also lead to a global reduction in H3K27 methylation, albeit with altered genomic distribution (Sarthy et al., 2020; Furth et al., 2022). Subsequently, H3K27M or EZHIP, a non-histone protein that functionally mimics H3K27M (Hübner et al., 2019; Sievers et al., 2021), were implicated in other cancers (Gessi et al., 2016; Koelsche et al., 2017; Lehnertz et al., 2017; Ryall et al., 2017; Pajtler et al., 2018; Hübner et al., 2019; Jain et al., 2019). Thus, H3K27M has dramatic effects on cell physiology. Mutating other highly conserved lysine residues to methionine along the H3 tail revealed that H3K36M and H3K9M also suppress methylation at their respective, specific residues (Lewis et al., 2013), whereas the effects of H3K4M are more complex. H3K36M was subsequently identified in chondroblastoma, soft-tissue sarcoma, and head and neck squamous cell carcinoma (Behjati et al., 2013; Lu et al., 2016; Papillon-Cavanagh et al., 2017). Likewise, H3K4M destabilizes tumor suppressors MLL3 (KMT2C) and MLL4 (KMT2D; Jang et al., 2019), and the mutation has been found in cancer (Nacev et al., 2019). Expression of H3K9M has not been prominently identified in human malignancies, but was associated with increased incidence of T-cell lymphomas in mice (Brumbaugh et al., 2019a). Thus, the term ‘oncohistones’ was coined to describe the various K-to-M mutations. Although the role of K-to-M mutants and other histone variants (Box 1) has been extensively studied in the context of cancer (Qiu et al., 2018; Nacev et al., 2019; Zhang and Zhang, 2019; Krug et al., 2021; Deshmukh et al., 2022; Mitchener and Muir, 2022), there is increasing interest in using K-to-M mutants as tools to study the role of specific histone modifications in diverse biological processes (Table 1) (Herz et al., 2014; Brumbaugh et al., 2019a). Here, we review the properties that make K-to-M mutants tractable systems for studying histone methylation and highlight their use in recent studies, with a focus on work in stem cells and development. We discuss important limitations to consider when using K-to-M mutants and interpreting experimental results. Overall, K-to-M mutants provide a system to directly manipulate histone methylation during development and, therefore, overcome many limitations described above.

K-to-M mutants represent a valuable tool to study histone methylation for many reasons (Table 2). Owing to their dominant-negative nature, K-to-M mutants permit manipulation of histone methylation without editing every histone H3 gene in the genome or perturbing multiple methyltransferases (Miller et al., 2010; Xu et al., 2012; Warrier et al., 2022). Because K-to-M mutants are genetically encoded, they can be expressed in a tissue- and time-specific manner using cell type-specific or inducible promoters (Herz et al., 2014; Stafford et al., 2018; Brumbaugh et al., 2019a; Chaouch et al., 2021). This is a particularly important advantage for studying development because it permits spatiotemporal control over chromatin states that regulate cell fate specification and is feasible with small numbers of cells (i.e. it can be used in early embryos). Finally, K-to-M mutants suppress, but do not fully eliminate, histone methylation, acting essentially as hypomorphs, which permits manipulating histone marks in cases where complete disruption is deleterious to viability (Mohammad et al., 2017).

When using K-to-M mutants as tools, several important limitations must be considered (Table 2). Although no pleiotropic effects have been reported, it remains possible that K-to-M mutants have unknown functions outside of suppressing histone methylation. Although subject to the caveats discussed above, orthogonal approaches, such as methyltransferase knockdown or inhibitor assays, can address this issue by confirming phenotypes consistent with K-to-M mutants (Lu et al., 2016; Brumbaugh et al., 2019a). Another caveat is that expressing dominant-negative K-to-M mutants affects histone methylation genome-wide. This feature makes it difficult to distinguish the contribution of histone methylation at a single gene or locus to the overall phenotype. In addition, K-to-M mutants affect mono- (me1), di- (me2) and tri- (me3) methylation (Lewis et al., 2013). Often, different methylation states have distinct genomic distributions and associations with gene expression. Alternative approaches are necessary to clarify contributions of specific methylation states to the regulation of gene expression during differentiation and development. For example, we and others have performed knockdown experiments for methyltransferases that catalyze specific methylation states to tease apart the respective roles of mono-, di- and tri-methyl marks (Lu et al., 2016; Zhuang et al., 2018; Brumbaugh et al., 2019a; Rajagopalan et al., 2021). Targeting specific methylation states may also be possible through methyltransferase-specific inhibitors (e.g. developing inhibitors for SETD2 would permit characterization of H3K36me3, as SETD2 is the only enzyme that deposits H3K36me3). In addition, interplay often exists between histone marks so that loss of one modification may affect the abundance or distribution of other histone modifications. In many cases, corresponding gene expression changes result from perturbation of both histone modifications. For example, H3K27M expression leads to loss of H3K27 methylation, but also changes in H3K27 acetylation (H3K27ac), which may influence transcription (Nagaraja et al., 2017; Piunti et al., 2017; Brien et al., 2021). Nucleosomes containing H3K27M were also reported to interact with MLL1 (KMT2A), leading to a redistribution of the activating H3K4me3 mark (Furth et al., 2022). Thus, it is important to consider the broad impact of K-to-M mutants on chromatin when interpreting experimental results. Likewise, an important consideration is the choice of histone variant used to express K-to-M mutants, as replication dependent and replication independent histone H3 variants are distributed differently across the genome and incorporated at different times in the cell cycle. Indeed, H3K27me3 profiles were distinct between systems expressing H3.1K27M and H3.3K27M (Sarthy et al., 2020). It is worth noting that because K-to-M mutants are often found on the replication-independent H3.3 variant in cancers, many of the constructs used to overexpress histone mutants are generated using H3.3. A final consideration is the mechanism of each K-to-M mutant. Current models suggest that H3K9M and H3K36M sequester respective methyltransferases as a means of inhibition, whereas the mechanism for H3K27M is more complex. Although more work is needed in this area, many methyltransferases have regulatory roles outside of methylation that may be affected by their sequestration. Overall, the primary barrier to using K-to-M mutants is that our understanding of their mechanism remains limited, which is important to consider when designing and interpreting experiments.

H3K27me2/3 are associated with transcriptional silencing, a regulatory mechanism that is conserved among animals, plants and fungi (Bemer and Grossniklaus, 2012; Lanzuolo and Orlando, 2012; Lewis, 2017). Polycomb repressive complex 2 (PRC2) is responsible for depositing H3K27 methylation, primarily within promoter regions, through the catalytic SET domain-containing subunit, EZH2 (Cao et al., 2002; Fischle et al., 2003; Min et al., 2003). For detailed information on enzymes that regulate H3K27 methylation, we direct the reader to comprehensive reviews (Margueron and Reinberg, 2011; Perino and Veenstra, 2016; Schuettengruber et al., 2017; Wiles and Selker, 2017).

In mammals, H3K27me3 at the promoters of developmental regulators is thought to direct lineage-specific transcription programs (O'Carroll et al., 2001; Mohn et al., 2008; Zhou et al., 2011). Consistent with this notion, developmental genes are derepressed following loss-of-function mutations in polycomb group proteins, inducing abnormal differentiation patterns and tissue specification (Margueron and Reinberg, 2011; Bemer and Grossniklaus, 2012; Lanzuolo and Orlando, 2012; Di Croce and Helin, 2013; Laugesen et al., 2016). Despite these associations, studying the direct function of H3K27 methylation in development is challenging. Although EZH2 is the primary enzyme that catalyzes H3K27me1/2/3, it has known regulatory functions that are independent of H3K27 methylation (Xu et al., 2012) and EZH2 knockout is embryonic lethal (O'Carroll et al., 2001; Herz et al., 2013). These features limit the effectiveness of approaches that seek to knockout or knockdown EZH2 and highlight the importance of H3K27M in studying H3K27 methylation, particularly in development.

Our mechanistic understanding of K-to-M mutant histone function is largely based on studies of H3K27M. Despite robustly suppressing H3K27 methylation, H3K27M mutant histones accounted for only 3-18% of total H3 in cancer cells (Lewis et al., 2013). These data suggest that H3K27M is a dominant-negative inhibitor of PRC2; however, the precise mechanism is likely complex (Deshmukh et al., 2022). Structural studies showed that the methionine substitution on H3K27M occupies the active site of EZH2 similar to the wild-type lysine residue (Fig. 2A) (Jiao and Liu, 2015; Justin et al., 2016), but H3K27M peptides exhibited higher affinity for EZH2 compared with wild-type H3 peptides (Justin et al., 2016). These findings led to a sequestration model in which EZH2 binds H3K27M but is unable to undergo a productive enzymatic reaction and remains bound with high affinity to the mutant substrate. However, several observations extended our mechanistic understanding of H3K27M. In contrast with peptide-based studies, binding assays using nucleosomes revealed only a modest difference in affinity between EZH2 and either wild-type or mutant histone (Wang et al., 2017). Moreover, chromatin profiling showed that, although expression of H3K27M led to global reduction in H3K27 methylation, high affinity PRC2 binding sites retained narrow, but detectible, regions of H3K27me3 (Bender et al., 2013; Chan et al., 2013a,b; Mohammad et al., 2017; Harutyunyan et al., 2019). These data suggest that H3K27M suppresses H3K27me3 spread more potently than initial deposition (Harutyunyan et al., 2019). Further, H3K27me2, which is also catalyzed by EZH2, was decreased but remained broadly distributed in H3K27M-expressing cells (Harutyunyan et al., 2019, 2020). Thus, residual EZH2 activity persists and is not restricted to specific genomic loci. Finally, multiple groups have reported that H3K27M and EZH2 do not necessarily colocalize on DNA (Piunti et al., 2017; Harutyunyan et al., 2019, 2020; Silveira et al., 2019; Sarthy et al., 2020). These observations suggest that PRC2 sequestration is not the sole mechanism underlying H3K27M inhibitory effect on H3K27 methylation. Recent work using an inducible histone mutant demonstrated that interaction between EZH2 and H3K27M is dynamic, with strong, transient colocalization after 6 h of H3K27M expression that diminishes after 24 h (Stafford et al., 2018). Corresponding in vitro assays revealed that H3K27M-mediated inhibition of EZH2 persisted, even after its release from mutant histone. In addition, the inhibitory effect of H3K27M is greater on the allosterically activated form of PRC2 (Stafford et al., 2018), which facilitates the spread of H3K27me3. These data are consistent with a revised model in which H3K27M binds EZH2 but does not necessarily restrict its localization. Instead, durable inhibition of EZH2, particularly for the allosterically activated form of PRC2, prevents spreading of H3K27me3 and causes a global decrease of the repressive mark.

Given its dominant-negative effect, H3K27M offers a valuable tool to assess the role of H3K27 methylation in differentiation and development (Herz et al., 2014; Piunti et al., 2017; Stafford et al., 2018; Jain et al., 2020; Chaouch et al., 2021). Tissue-specific induction of H3K27M in Drosophila wing imaginal discs strongly reduced H3K27 methylation and phenocopied developmental defects observed with PRC2 loss-of-function mutations (Herz et al., 2014). Correspondingly, PRC2-target genes important to development, such as Ubx and wg, were aberrantly upregulated, suggesting a functional role for H3K27 methylation in suppressing lineage-inappropriate gene expression (Herz et al., 2014). Likewise, conditional expression of H3K27M from the Drosophila eyeless promoter led to abnormal upregulation of developmental genes, including the PRC2-regulated Hox gene Antp, in anterior compartments of flies (Chaouch et al., 2021). As a consequence, cell polarity was disrupted in eye imaginal discs of H3K27M-expressing flies (Chaouch et al., 2021). Unexpectedly, genes related to photoreceptor cell differentiation were downregulated following H3K27M expression and adult flies presented with hypotrophic, disorganized eyes that contained fewer photoreceptor neurons (Chaouch et al., 2021). The authors noted that expression of H3K36M induced a similar phenotype (discussed below), suggesting interplay between antagonistic H3K27 and H3K36 methyl marks (Chaouch et al., 2021). Indeed, redistribution of H3K36me2, a histone modification that is generally associated with active transcription, followed loss of H3K27 methylation (Chaouch et al., 2021). These studies underscore the importance of regulatory crosstalk between chromatin modifications and highlight the power of using H3K27M in tissue-specific systems to study the role of H3K27 methylation in developmental programs.

Histone modifications are thought to contribute to development by establishing and ultimately maintaining cell type-specific gene expression to safeguard cell identity (Brumbaugh et al., 2019b). However, cell fate may be altered experimentally through a process called reprogramming (Takahashi and Yamanaka, 2006; Takahashi et al., 2007; Yu et al., 2007) or in vivo through dedifferentiation (Katsuyama and Paro, 2011; Saxena and Shivdasani, 2021). How histone modifications influence these processes is a complex and unresolved question. To investigate the role of H3K27 methylation in reprogramming, researchers co-induced expression of H3K27M and the transcriptional activator Vestigial (Vg) in Drosophila eye imaginal discs (Ahmad and Henikoff, 2021). Vg is a master regulator of wing development and its ectopic expression is sufficient to reprogram tissue into wing structures (Kim et al., 1996). However, when H3K27M was expressed simultaneously with Vg in eye imaginal discs, eye-to-wing reprogramming was undetectable (Ahmad and Henikoff, 2021). Reprogramming requires silencing of genes that maintain the original cell fate. These data suggest that suppressing H3K27 methylation prevents repression of these genes and disrupts reprogramming. Whether other K-to-M mutants similarly interfere with reprogramming warrants further study. Likewise, it will be interesting to test the effect of H3K27M in different cell types and using different reprogramming systems (i.e. reprogramming to pluripotency in mouse or human cells).

Analogous to reprogramming, tissue regeneration is associated with changes to cell plasticity and function (Katsuyama and Paro, 2011). For example, zebrafish are capable of cardiac tissue regeneration through cardiomyocyte dedifferentiation following injury (Jopling et al., 2010; Kikuchi et al., 2010). Chromatin modifications are likely involved in this process, as EZH2 is upregulated following cardiac injury in zebrafish. H3K27me3 correspondingly increased over sarcomeric and cytoskeletal genes that were downregulated in response to myocardial injury (Ben-Yair et al., 2019). To demonstrate the direct function of H3K27 methylation in cardiac regeneration, researchers induced H3K27M expression in zebrafish cardiomyocytes following injury. In comparison with wild-type H3 controls, cardiac muscle harvested from H3K27M-expressing fish had lower H3K27me3, was deficient in muscle renewal and showed variable but increased scarring following injury (Ben-Yair et al., 2019). Mechanistically, cardiomyocytes from fish expressing H3K27M failed to properly downregulate sarcomeric genes, likely leading to cell motility defects and decreased wound infiltration (Ben-Yair et al., 2019). In adult mammals, cardiomyocyte turnover and cardiac tissue regeneration decrease exponentially with age (Bergmann et al., 2015); more work is needed to determine whether suppressing mammalian homologs of the structural genes repressed by H3K27 methylation in zebrafish cardiomyocytes bestows regenerative capacity in mammals.

H3K27 methylation dramatically increases during differentiation, suggesting a functional role for the histone modification in development (Juan et al., 2016). To test this possibility, researchers expressed H3K27M in human embryonic stem cells. Suppressing H3K27 methylation led to marked differentiation defects relative to wild-type H3 control and increased clonogenicity under self-renewing conditions (Kfoury-Beaumont et al., 2022). These findings suggest that loss of H3K27 methylation maintains cells in a self-renewing, developmentally primitive state (Kfoury-Beaumont et al., 2022). Consistent with this observation, H3K27M expression in human P5K neural progenitor cells blocked astrocyte differentiation and induced gene expression that resembles a primitive, stem cell state (Funato et al., 2014). These studies provide functional evidence that H3K27 methylation is important during differentiation. An interesting remaining question is whether suppressing H3K27 methylation may facilitate cell fate transitions (e.g. transdifferentiation and reprogramming) by inducing a plastic cell state.

The effect of suppressing H3K36 methylation on gene expression is complex. Both H3K36me2 and H3K36me3 antagonize PRC2 (Schmitges et al., 2011; Yuan et al., 2011; Jani et al., 2019) and are thought to act as a barrier to the repressive H3K27me3 mark. Indeed, H3K27me3 is dramatically redistributed upon induction of H3K36M (Lu et al., 2016; Brumbaugh et al., 2019a; Chaouch et al., 2021; Hoetker et al., 2023), which likely influences transcription programs. H3K36me2/3 have also been implicated in regulation of transcriptional elongation, alternative splicing, and DNA replication and repair (Wagner and Carpenter, 2012; Wen et al., 2014; McDaniel and Strahl, 2017). H3K36me2 is often found in intergenic regions, whereas H3K36me3 usually marks gene bodies, but both modifications can be added and removed quickly by several methyltransferases and demethylases (Fig. 1) (Wagner and Carpenter, 2012). In mammals, NSD1/2/3 and SETD2 are considered the primary H3K36 methyltransferases. SETD2 is the only enzyme that deposits H3K36me1/2/3; the other methyltransferases catalyze H3K36me1/2 (Edmunds et al., 2008). For detailed information on enzymes that modify H3K36, we direct the reader to comprehensive reviews (Wagner and Carpenter, 2012; Hyun et al., 2017; Huang and Zhu, 2018).

H3K36M is thought to inhibit methylation through sequestration of cognate methyltransferases. Crystal structures show that insertion of the H3K36M side chain into the active site of SETD2 forms preferential contacts that stabilize their interaction (Fig. 2B) (Yang et al., 2016; Zhang et al., 2017). Unlike the other K-to-M mutants, this interaction is likely not dependent on the presence of the cofactor S-adenosylmethionine (SAM, the methyl donor), which suggests that the inhibitory effect of H3K36M is not necessarily due to non-productive catalysis (Zhang et al., 2017). Interestingly, although the H3K36 methyltransferases NSD1, NSD2 and SETD2 showed increased affinity for H3K36M compared with wild-type H3 under some conditions (Lu et al., 2016), NSD1 is not inhibited by the mutant histone in vitro (Fang et al., 2016; Zhang et al., 2017). Considering the sequence and structural similarities between NSD1 and NSD2, it will be interesting to probe the molecular differences between the proteins that allow NSD1 to evade inhibition. H3K36M suppresses both H3K36me2 and H3K36me3 (Fang et al., 2016; Lu et al., 2016; Brumbaugh et al., 2019a), though the effect appears to be context-dependent, as some studies report predominant loss of H3K36me2 (Zhuang et al., 2018; Yuan et al., 2020; Yano et al., 2022).

Given the general association between H3K36me2/3 and active transcription, H3K36M might be expected to broadly disrupt gene expression; however, we and others have observed that H3K36M induction affects only subsets of genes in different cell types (Brumbaugh et al., 2019a; Chaouch et al., 2021). These genes generally reflect cell fate or function, and their disruption may explain distinct differentiation defects observed in H3K36M systems. For example, expression of H3K36M in mesenchymal stem cells or preadipocytes strongly inhibited differentiation into mature adipocytes (Lu et al., 2016; Zhuang et al., 2018). Mechanistically, cells expressing H3K36M failed to upregulate adipocyte maturation genes, which correspondingly lost H3K36me2 and gained H3K27me3 (Lu et al., 2016; Zhuang et al., 2018). In vivo, expression of H3K36M from the adipose-specific Fabp4 promoter led to increased white adipose tissue and decreased brown adipose tissue. Transcriptional profiling revealed a concomitant increase in gene expression characteristic of white adipose tissue and decreased expression of brown adipose tissue markers. Compared with wild-type controls, however, total adipose tissue mass did not change in H3K36M-expressing mice, suggesting that white adipose tissue expanded in place of brown adipose tissue, or brown adipose tissue directly converted (i.e. reprogrammed) into white adipose tissue (Zhuang et al., 2018).These results suggest that H3K36 methylation affects the differentiation trajectory of adipose cell types. It is possible that master regulators of cell fate are dependent on H3K36 methylation to reinforce their expression, which may explain this dependence. Collectively, these experiments support the role of H3K36 methylation in adipogenesis and illustrate the significance of H3K36M in exploring chromatin-based regulation of different cell types.

Similar to adipose, regenerative tissues rely on chromatin-based regulation to direct differentiation and homeostasis (Yu et al., 2016). Correspondingly, H3K36M expression in adult mice revealed stage-specific roles for H3K36 methylation in the differentiation of blood, intestine and sperm (Brumbaugh et al., 2019a). RNA-seq analysis from the testes of male mice expressing H3K36M showed a specific downregulation of mature spermatocyte-associated genes, but not undifferentiated spermatogonia-related genes. These findings support a role for H3K36 methylation in regulating sperm maturation (Brumbaugh et al., 2019a). Likewise, suppression of H3K36 methylation disrupted differentiation at distinct stages in many hematopoietic lineages (Brumbaugh et al., 2019a). For example, H3K36M-expressing hematopoietic stem cells, erythrocytes, megakaryocytes and lymphocytes exhibited stage-specific differentiation defects. Genomic analyses from H3K36M-expressing hematopoietic stem and progenitor cells revealed a global loss of H3K36 methylation and corresponding downregulation of genes involved in specification of many of the affected lineages. These data provide a mechanistic basis for the observed differentiation defects; however, a subset of hematopoietic genes was aberrantly upregulated in H3K36M-expressing cells. Profiling H3K27me3, a mark that is antagonized by H3K36me2/3 (Jani et al., 2019), revealed substantial redistribution of H3K27me3 in cells expressing H3K36M, with decreased occupancy at upregulated genes (Brumbaugh et al., 2019a). Together, these data underscore the importance of considering crosstalk between histone modifications when resolving mechanistic and phenotypic results. Finally, discontinuing expression of H3K36M in an inducible mouse model revealed that H3K36me3 recovered after just 7 days. At the phenotypic level, H3K36M withdrawal rescued hematopoietic differentiation defects, suggesting that the effects of H3K36M expression are reversible (Brumbaugh et al., 2019a). These observations highlight the potential for using K-to-M mutants to study how histone modifications are established during differentiation and their importance in these processes.

H3K36M expression suppresses osteogenic and chondrogenic transcription programs (Behjati et al., 2013; Fang et al., 2016; Lu et al., 2016). These data suggest an essential role for H3K36 methylation in generating bone and cartilage. Indeed, limb-specific induction of H3K36M in chondrogenic progenitor cells led to shortened limbs in mice (Abe et al., 2021). On a molecular level, H3K36me2 decreased in H3K36M tibial midshafts and, correspondingly, terminal hypertrophic chondrocyte markers were suppressed in H3K36M samples (Abe et al., 2021). Based on co-immunoprecipitation experiments in chondrocytes, H3K36M preferentially interacted with NSD1 and NSD2, but not SETD2, suggesting that H3K36me2 functions in chondrocyte differentiation (Abe et al., 2021). Supporting this idea, knockout of NSD1/2, but not SETD2, phenocopied the effects of H3K36M on differentiation in chondrocytes (Rajagopalan et al., 2021). Further work is needed to determine the developmental state of H3K36M-expressing chondrocytes, as gene expression analysis suggested that the cells were not simply undifferentiated. It is possible that the unusual transcriptional profile induced by H3K36M occurs in other tissues or mimics a different developmental stage.

H3K36 methylation has also been implicated in regulating DNA methylation in early development and gametogenesis (Choufani et al., 2015; Papillon-Cavanagh et al., 2017; Weinberg et al., 2019; Xu et al., 2019; Shirane et al., 2020; Yano et al., 2022). This interaction is likely to be important in mammalian oocytes, which have unique DNA methylation landscapes that are necessary for proper reproduction (Yano et al., 2022). To better understand the interplay between H3K36 methylation and DNA methylation, researchers expressed H3K36M under an oocyte-specific promoter in mice (Yano et al., 2022). Loss of H3K36me2 in H3K36M-expressing oocytes correlated with decreased CpG methylation, but gene expression at these loci was largely unaffected. These data suggest that H3K36me2-directed DNA methylation is not necessary for regulating transcription in oocytes. However, following fertilization, embryos derived from H3K36M oocytes failed to progress past the peri-implantation stage, suggesting that H3K36M either durably alters the chromatin landscape of oocytes in a way that is incompatible with development or H3K36M persists within the embryo, though undetectable at the blastocyst stage (Yano et al., 2022). Collectively, these experiments provide a basis for H3K36me2 regulation of DNA methylation in oocytes and illustrate the utility of H3K36M in studying chromatin-based regulation in embryogenesis.

As a tool, H3K36M expression is tractable in diverse model systems and was used to study the role of histone methylation in determining cell fate in Drosophila (Herz et al., 2014; Chaouch et al., 2021). Specifically, expression of H3K36M in eye imaginal discs led to disorganized eye structures and overall eye hypotrophy (Chaouch et al., 2021). This tissue showed a loss of both H3K36me2 and H3K36me3 at canonical eye specification genes, as well as an increase of antagonistic H3K27me3 in the H3K36M background. Moreover, knocking down E[z], the Drosophila homolog of EZH2, rescued H3K36M-associated phenotypes, again suggesting that finely-tuned cross-talk between these marks is needed to maintain proper gene expression and repression (Chaouch et al., 2021). Curiously, loss of H3K36me2 over pericentromeric regions in Drosophila led to derepression of transposable elements. These results suggest that H3K36me2 may play a repressive role under some conditions. Supporting this notion, H3K36M-mediated loss of H3K36me2 relieved enhancer silencing in mouse embryonic fibroblasts undergoing reprogramming, leading to efficient upregulation of pluripotency-related genes (Hoetker et al., 2023).

Though the mechanism of K-to-M mutants remains an area of active investigation, H3K36M has been used to great effect to address the role of H3K36 methylation in developmental processes. Often, knocking down a single methyltransferase does not fully replicate phenotypes observed when expressing H3K36M, highlighting the utility of this dominant-negative mutant. Moreover, H3K36me2 and H3K36me3 both appear to regulate developmental processes, but the contributions of each mark are likely cell type- and context-specific (Fang et al., 2016; Lu et al., 2016; Brumbaugh et al., 2019a; Weinberg et al., 2019; Chaouch et al., 2021; Yano et al., 2022). Outside of development, H3K36M has also been used to study the role of H3K36 methylation in the DNA damage response (Pfister et al., 2014) and viral lifecycle of human papilloma virus (Gautam et al., 2018).

H3K9 methylation is generally associated with transcriptional silencing (Schotta et al., 2002; Nicetto and Zaret, 2019), though each methylation state has a distinct role and distribution in the genome. H3K9me1 localizes to silenced genes in euchromatic regions, although it is also enriched at transcription start sites of actively transcribed genes in mammalian cells (Barski et al., 2007; Gupta et al., 2012). By contrast, H3K9me2 marks peripheral heterochromatin (Poleshko et al., 2019) and is enriched in lamina-associated domains (Fukuda et al., 2021), whereas H3K9me3 localizes to transposable elements across the genome (Mikkelsen et al., 2007; Leung and Lorincz, 2012; Bulut-Karslioglu et al., 2014) and silences lineage-inappropriate genes during differentiation (Nicetto and Zaret, 2019). Both H3K9me2 and H3K9me3 are enriched at centromeres and telomeric regions, as well as in non-coding repetitive regions, thus defining constitutive heterochromatin (Ninova et al., 2019; Padeken et al., 2022). These marks are deposited by several methyltransferases (Fig. 1) that can establish different methylation states. In yeast for example, Clr4 deposits all three forms of H3K9 methylation, whereas in mammalian cells, EHMT1 (GLP) and EHMT2 (G9a) are responsible for both mono- and di-methylation (Nicetto and Zaret, 2019). In addition, SUV39H1, SUV39H2 and SETDB1 are capable of catalyzing mono-, di- and tri-methylation (Hyun et al., 2017). For detailed information on the role of H3K9 methylation in gene silencing and heterochromatin maintenance, we refer the reader to comprehensive reviews (Shinkai and Tachibana, 2011; Hyun et al., 2017; Nicetto and Zaret, 2019; Ninova et al., 2019).

Structural analyses of H3K9M revealed that the mutant histone acts as an orthosteric inhibitor of the H3K9 methyltransferase, EHMT2 (Jayaram et al., 2016; Shan et al., 2016) (Fig. 2C). Positioning of the H3K9M methionine side chain suggested increased van der Waals forces compared with wild-type H3, a finding that is reinforced by the observation that EHMT2 bound with higher affinity to H3K9M peptides compared with wild-type H3 peptides (Shan et al., 2016). Whether H3K9M-containing nucleosomes also demonstrate higher affinity for H3K9 methyltransferases remains an important but unresolved question. In yeast, H3K9M led to an accumulation of the H3K9 methyltransferase, Clr4, at heterochromatin nucleation centers (Shan et al., 2016). Collectively, these data support a sequestration model in which H3K9M methyltransferases lock on to the H3K9M substrate, preventing modification of wild-type H3. However, nuanced work is needed to fully understand the mechanism of H3K9M. Similar to H3K27M, expressing H3K9M in yeast prevented spread of H3K9me3 to regions surrounding heterochromatin nucleation centers (Shan et al., 2016). Moreover, H3K9M-containing nucleosomes outside of heterochromatin nucleation centers do not colocalize with Clr4 (Shan et al., 2016), suggesting that association between H3K9M and Clr4 is regulated by other means. This observation raises the possibility that a complex mechanism, similar to H3K27M, may be the basis for the H3K9M inhibitory effect.

H3K9M has been used to study development processes in several model organisms. In Drosophila, H3K9 methyltransferase Su(var)3-9 promotes heterochromatin formation and silencing of nearby euchromatic genes via position effect variegation (PEV) (Schotta et al., 2002). Until recently, it was challenging to directly test the role of H3K9 methylation in PEV, so researchers generated an inducible system to express H3K9M in flies (Herz et al., 2014). Inducing H3K9M globally reduced H3K9 methylation and suppressed PEV in eye imaginal discs and salivary glands (Herz et al., 2014). Mechanistically, co-immunoprecipitation mass spectrometry studies revealed reduced association between H3K9M-containing mono-nucleosomes and HP1 proteins, which are recruited to chromatin by H3K9 methylation to mediate chromatin compaction and transcriptional silencing (Herz et al., 2014). Together, these data provide evidence that Su(var)3-9 regulates PEV directly through H3K9 methylation.

In mammals, H3K9 methylation and corresponding modifying enzymes are widely implicated in cell fate and development (Sridharan et al., 2013; Becker et al., 2016; Nicetto and Zaret, 2019; Ninova et al., 2019). To better understand the direct role of H3K9 methylation in differentiation, we recently developed a doxycycline-inducible H3K9M mouse model (Brumbaugh et al., 2019a). Expressing H3K9M in embryonic stem cells from this model led to a pronounced differentiation defect in embryoid body assays. Correspondingly, ATAC-seq analyses demonstrated that chromatin remained accessible at key genes related to pluripotency (e.g. Nanog, Oct4 and Sox2) after 3 days of differentiation in H3K9M-expressing cells. These observations suggest that decreased H3K9 methylation prevents chromatin compaction and silencing of genes that maintain self-renewal in embryonic stem cells. Using the same model in vivo, we found that inducing H3K9M in adult mice led to global loss of H3K9me3 and defects in megakaryocyte, proB cell, and hematopoietic stem and progenitor cell differentiation. Genes related to immature and progenitor cells were upregulated in H3K9M cells, in line with the repressive role of H3K9 methylation. Overall, these studies indicate that H3K9 methylation plays an important role in suppressing stem cell maintenance genes during differentiation in both embryonic and adult stem cells.

H3K9 methylation has also been implicated in the development of specialized organ systems, such as the heart. For example, cardiomyocyte maturation triggers dramatic changes to three-dimensional nuclear organization and a striking redistribution of heterochromatin from the center of the nucleus to the nuclear periphery (Seelbinder et al., 2021). To determine whether H3K9me3 contributes to cardiomyocyte nuclear reorganization, researchers induced expression of H3K9M in mouse embryonic cardiomyocytes (Seelbinder et al., 2021). After 4 days, H3K9M expression disrupted overall chromatin reorganization and abrogated H3K9me3 accumulation at the nuclear periphery. RNA-seq analysis revealed that genes involved in cardiac development were significantly decreased in H3K9M-expressing cardiomyocytes, suggesting that H3K9 methylation plays a functional role in differentiation. However, although global gene expression decreased in H3K9M-expressing cells, proximity of those genes to H3K9me3 domains did not correlate with expression-level differences. These data suggest that H3K9me3 contributes to gene regulation indirectly, perhaps through nuclear organization, rather than by directly controlling transcription.

Methylation of H3K4 is generally associated with euchromatin and actively transcribed DNA (Zhou et al., 2011). H3K4 methylation is predominantly found at enhancers and promoters, and the H3K4 landscape is often distinct between cells, suggesting a role for H3K4 methylation in regulating cell type-specific gene expression profiles (Heintzman et al., 2007, 2009). However, regulation associated with H3K4 methylation is more nuanced and is influenced by the genomic location of the mark and its association with other histone modifications. For example, although H3K4me1 is found at active enhancers, the presence or absence of H3K27ac indicates whether the enhancer is active or poised, respectively (Creyghton et al., 2010). The retention of H3K4me1 in the absence of H3K27ac resulted in faster activation of affected genes, demonstrating that H3K4 methylation also plays a role in chromatin-mediated memory (Ostuni et al., 2013). Similarly, H3K4me3 typically marks active promoters (Heintzman et al., 2007); however, it is also found at the transcription start sites of developmentally suppressed genes in embryonic stem cells, together with the repressive H3K27me3 mark (Bernstein et al., 2006). These bivalent domains are thought to represent a poised chromatin state that resolves into a single mark during differentiation to suppress or activate expression of developmentally controlled genes (Bernstein et al., 2006). Thus, by coordinating gene activation, H3K4 methylation plays an important role in differentiation and maintenance of cell states.

H3K4 methyltransferases are highly conserved across species. In yeast, the Set1 methyltransferase deposits all H3K4 methyl marks and serves as the core of the multi-subunit COMPASS complex (Briggs et al., 2001; Miller et al., 2001; Roguev et al., 2001). There are six Set1 homologs in mammals: SET1A, SET1B, MLL1, MLL2 (KMT2B), MLL3 and MLL4. The homologs function as scaffold proteins for SET1/MLL family protein complexes, which share common subunits but also contain unique proteins that direct specific functions. For detailed information, we direct the reader to comprehensive reviews (Shilatifard, 2012; Hyun et al., 2017). In addition, SET7/9, SMYD1-3, SETMAR and PRDM9 catalyze H3K4 methylation based on biochemical assays (Wang et al., 2001; Nishioka et al., 2002; Hamamoto et al., 2004; Lee et al., 2005; Tan et al., 2006; Abu-Farha et al., 2008; Sirinupong et al., 2010; Gu and Lee, 2013; Spellmon et al., 2015; Powers et al., 2016).

Contrasting other K-to-M mutants, the reported effects of H3K4M on H3K4 methylation vary and may be context-dependent. Whereas some studies observed full reduction of H3K4 methylation marks (Jang et al., 2017, 2019), others reported no substantive change in methylation (Lewis et al., 2013). Unlike other K-to-M mutants, several H3K4 methyltransferases do not have a higher binding affinity for H3K4M (Burton et al., 2020), arguing against a sequestration-based mechanism. Instead, expression of H3K4M resulted in lower levels of MLL3 and MLL4 protein, but not mRNA, suggesting that the mutant histone directly destabilizes some H3K4 methyltransferases (MLL1 and MLL2 were unaffected) (Jang et al., 2017, 2019). Moreover, several studies found that depletion of H3K4 methylation required multiple replication cycles (Chan et al., 2013a,b; Jang et al., 2019). It is tempting to speculate that H3K4M prevents H3K4 methylation by destabilizing methyltransferases, whereas existing methylation is progressively diluted as cells divide, rather than actively removed by demethylases. Supporting this point, H3K4M has a strong binding affinity to LSD1 (KDM1A) and LSD2 (KDM1B), both of which demethylate H3K4me1/2 (Forneris et al., 2007; Chen et al., 2013; Zhang et al., 2013), and structural analyses are consistent with these observations (Fig. 2D) (Forneris et al., 2007). By suppressing demethylases, H3K4M may slow or prevent H3K4 methylation loss in certain contexts. Collectively, these data may resolve the variable effects of H3K4M on histone methylation, which likely depends on the levels and activity of methyltransferases and demethylases present in a given cell, as well as cell proliferation rate and time since induction of H3K4M. Like other K-to-M mutants, H3K4M-induced depletion of H3K4 methylation affects other histone modifications (Jang et al., 2019). This includes depletion of H3K27 acetylation, a mark of active enhancers that is often found in association with H3K4me1 (Jang et al., 2019). Thus, the mechanism underlying H3K4M function is complex and warrants careful consideration when interpreting experimental data.

H3K4M has been used to study the role of H3K4 methylation during development and differentiation. For example, MLL3 and MLL4, the primary writers of H3K4me1 and H3K4me2, are required for adipose cell fate transitions, based on knockout studies (Lee et al., 2013). To clarify this point using an orthogonal system, researchers expressed H3K4M in pre-adipocytes during differentiation in culture, which reduced H3K4me1/2 and correspondingly, H3K27ac (Jang et al., 2019). As a result, adipose differentiation decreased, but cell proliferation was unaffected (Jang et al., 2019). In vivo induction of H3K4M reduced differentiation of adipose and muscle cells but did not affect maintenance or function of these tissues, suggesting that H3K4 methylation is primarily needed for cell fate decisions in this system (Jang et al., 2019). This is in line with previous work, which showed that MLL3 and MLL4 were not involved in adipose cell maintenance, nor were they involved in maintenance of active enhancers (Wang et al., 2016). Instead, H3K4M disrupted enhancer activation at genes required for adipose differentiation and subsequently reduced expression of pro-differentiation genes (Jang et al., 2019). This study highlights the importance of H3K4 methylation in enhancer activation, specifically during differentiation (Jang et al., 2019). Notably, perturbations to H3K4 methylation affect differentiation in different contexts, suggesting that H3K4 methylation may be more generally required for cell fate transitions (Lee et al., 2013; Wang et al., 2016; Froimchuk et al., 2017). Moving forward, H3K4M mutations may be useful in clarifying these relationships.

Studying chromatin during embryogenesis is challenging because a limited number of cells are available. However, H3K4M can be conditionally expressed at any developmental time point and is therefore a tractable tool to study the role of H3K4 methylation in early development. One example is the role of H3K4 methylation in zygotic genome activation. Previous research showed that maternal and paternal pronuclei activate transcription asymmetrically (Aoki et al., 1997), potentially resulting from chromatin-level differences (Aoshima et al., 2015). To directly test this hypothesis, H3K4M mRNA was injected into unfertilized oocytes (Aoshima et al., 2015). As a result, H3K4me1 and H3K4me3 decreased in the paternal pronucleus but were unaffected in the maternal pronucleus. Correspondingly, H3K4M was only observed in the paternal pronucleus, suggesting that incorporation of H3K4M into chromatin is necessary for its suppressive effect in the zygote. Functionally, loss of H3K4 methylation led to embryonic growth arrest at the eight-cell stage. However, the basis for the developmental defect occurred earlier in development, likely during the minor wave of zygotic genome activation, because injection of H3K4M mRNA after fertilization had no effect on embryo viability (Aoshima et al., 2015). Knocking down MLL3/4 phenocopied H3K4M induction, confirming that H3K4 methylation is crucial for minor zygotic genome activation in the paternal pro-nucleus (Aoshima et al., 2015). An interesting observation from this study is that incorporation of the K-to-M mutant histones into chromatin depended on the H3 variant injected into zygote-stage embryos. Whereas H3.3K4M reduced global H3K4 methylation in the paternal pronucleus, H3.1K4M had no effect. This finding contrasts observations in 293 T cells (Aoshima et al., 2015), where H3.1K4M and H3.3K4M both reduced H3K4 methylation, and underscores the importance of considering histone variant and cell type when using K-to-M mutants.

Collectively, these studies show that H3K4M is a useful tool to study development, provided that histone variant, expression timing, induction duration and cell type are considered. Nonetheless, H3K4M can address nuanced and challenging aspects of development and stem cell biology. For example, expressing H3K4M in embryonic stem cells during differentiation could define the function and importance of H3K4me3/H3K27me3 bivalent domains. Likewise, suppressing H3K4 methylation during different stages of hematopoiesis could clarify the role of H3K4 methylation for lineage commitment and maintenance. Indeed, researchers used H3K4M to confirm a role for H3K4me3 in genome stability (Mishra et al., 2018), underscoring the versatility of this tool in different contexts.

During development and homeostasis, cell identity is established and maintained by distinct gene expression programs (Zhao et al., 2017). Gene expression is directed, in part, by cis-acting elements (e.g. gene promoters and enhancers) that are regulated by chromatin modifications, including histone methylation (Heintzman et al., 2007; Zhao et al., 2017). These points underscore the importance of understanding chromatin-based regulation in development; however, direct, functional assays to determine the role of histone modifications are challenging. To meet this need, histone H3 K-to-M mutants have emerged as a versatile system to identify relationships between developmental transitions and histone methylation. Because K-to-M mutants act as dominant-negative inhibitors and are genetically encoded, it is possible, in principle, to target any developmental time point or tissue for analysis. For example, a doxycycline-inducible H3K9M mouse model (Brumbaugh et al., 2019a) can be made tissue-specific by using an rtTA transactivator that is driven from a neuron-specific promoter. The functional role of H3K9 methylation in neural development is then readily tested by administering doxycycline at a developmental time point of interest. In this way, K-to-M mutants provide both spatial and temporal control over specific chromatin states.

A further advantage of inducible K-to-M mutant systems is the ability to withdraw mutant histone expression and restore histone methylation (Brumbaugh et al., 2019a). Following suppression of methylation, this feature can be used to study the re-establishment of a particular histone mark and determine how restoring the mark affects cell fate decisions. Illustrating this point, H3K27 methylation is known to spread from nucleation points that overlap with developmental genes (Kraft et al., 2022). It will be interesting to test whether H3K27me3 properly spreads from these nucleation points following withdrawal of H3K27M and how the dynamics of this mark influences the expression of developmental genes and correspondingly, cell fate decisions. Notably, these kinds of analyses are not possible using knockout approaches for histone methyltransferases without additional genetic manipulation. Thus, K-to-M mutants are uniquely suited for this purpose.

Several studies have shown that expressing K-to-M mutants in stem or progenitor cells disrupts differentiation and maintains a plastic, developmental state (Funato et al., 2014; Lu et al., 2016; Papillon-Cavanagh et al., 2017; Brumbaugh et al., 2019a). An interesting and unanswered question is whether K-to-M mutants are capable of inducing plasticity in differentiated cells, particularly in the context of regeneration. Given the effects of K-to-M mutants on differentiation, it is tempting to speculate that suppressing histone methylation following injury could facilitate dedifferentiation to improve tissue repair. Indeed, during the preparation of this paper, researchers demonstrated that expression of H3K36M dramatically improved reprogramming efficiency (Hoetker et al., 2023). Overall, the utility of K-to-M mutants extends to many biological processes (Table 1), though key limitations must be considered when designing and interpreting experiments that use them (Table 2). We anticipate that K-to-M mutants will expand our understanding of direct functional roles of chromatin in many biological contexts.

We thank Peter Dempsey and Patricia Ernst for feedback and discussion. We apologize to the groups whose studies we did not have space to discuss.