Histone post-translational modifications (PTMs) constitute a major epigenetic mechanism for the control of gene expression. Histone marks have been detected on various residues, located either within the histone globular domain or along the tail, where they can affect the condensation, packaging, or binding of proteins to chromatin, which intricately regulate processes from gene expression to genomic stability. Due to this functional importance, aberrant patterns of histone PTMs have been implicated in various diseases including cancer and cardiovascular disease (Bannister and Kouzarides, 2011; Abi Khalil, 2014).

Owing to the advancement of high-sensitivity mass spectrometry, that has emerged as the gold standard technique for the identification of novel protein modifications, this has greatly expanded the catalogue of known histone PTMs. This includes the identification of a group of “short chain Lys acylations” that include Lys butyrylation, propionylation (Chen et al., 2007), formylation (Jiang et al., 2007), succinylation, malonylation (Xie et al., 2012), 2-hydroxyisobutyrylation (Dai et al., 2014), b-hydroxybutyrylation (Xie et al., 2016), glutarylation (Tan M. et al., 2014), benzoylation (Huang et al., 2018) and crotonylation (Kcr) (Tan et al., 2011). These modifications are similar to the archetypal Lys acetylation (Kac), but differ in hydrocarbon chain length, hydrophobicity or charge (Table 1). Mounting evidence suggests that these new histone marks can affect gene regulation and are functionally distinguishable from the commonly studied histone Kac, adding another level of complexity to chromatin biology (Sabari et al., 2017).

Histone Kcr was first identified in 2011 where it was found to be mainly associated with active chromatin (Tan et al., 2011). Since then, there has been growing interest in this modification as it has emerged as a powerful novel epigenetic mark. Like acetylation, crotonylation also occurs on the ε-amino group of Lys residues and modifies histone charge. Also similar to acetylation, the substrate for crotonylation is a donor molecule linked by a thioester to the sulfhydryl group of coenzyme A (CoA), namely crotonyl-CoA. A key question that has emerged following the discovery of Kcr, as with other newly identified acylations, is whether Kcr is functionally redundant from histone Kac or if it has a distinct role in regulating gene function. Here, the discovery and functional characterisation of histone crotonylation is described, as well as recently identified histone Kcr “writers,” the enzymes that catalyse this covalent modification, “erasers,” the enzymes that remove this modification and “readers,” the effector proteins that bind to histones in a crotonylation-dependent manner. In addition to the enzymatic regulation of histone crotonylation, the impact of cellular metabolism on this epigenetic process are also discussed. Finally, recent advances into the role of histone Kcr in health and disease are described.

When histone Kcr was first identified, it was found to be evolutionary conserved from yeast to human, occurring broadly in all core histones (H2A, H2B, H3, and H4), as well as linker histone H1 and marked active promoters and potential enhancers (Tan et al., 2011). Similar to Kac, Kcr also occurs on the ε-amino group of the lysine side chain, where it neutralizes the positive charge of this residue. The loss in positive charge on histone Lys residues weakens DNA interaction, thus making chromatin less compact and accessible to DNA-binding factors. In support of a potential cis-function of Kcr on chromatin structure, H3K122cr-H4 containing tetrasomes that were subjected to thermal stability assays, were found to be less stable compared to unmodified H3-H4 tetrasomes (Suzuki et al., 2016). Consistent with this, the ability of Kcr to destabilise nucleosome structure has been proposed to be part of a compensatory mechanism during chromatin-to-nucleoprotamine transition, an essential process during spermatogenesis as discussed in the spermatogenesis section below (Montellier et al., 2013). Montellier et al. (2013) showed that incorporation of a histone H2B variant, TH2B, is essential for the final transformation of dissociating nucleosomes into protamine packed structures. In the absence of TH2B, cells compensate by upregulating H2B and programming nucleosome instability to reach that of wild type cells through targeted histone modifications, including crotonylation of H3K122 and H4K77. This in turn allows the histone replacement to take place. Furthermore, modified histone lysine residues can mediate trans-effects through recruitment of effector proteins containing specific reader modules. This is particularly important for histone Kcr, where the crotonyl group is a four-carbon chain containing a C–C π bond that results in a rigid planar conformation, which is unique among histone acylations. The extended hydrocarbon chain of the crotonyl group increases the hydrophobicity and bulk of the Lys residue compared to acetylation (Sabari et al., 2017). These differences in the biophysical properties of the crotonyl group provide an important mechanism of specificity for reader interaction, as described in detail below.

Histone Kcr is dynamically regulated by the opposing enzymatic activities of writers and erasers. To date, no selective enzymes that directly add or remove crotonyl groups from modified lysine residues have been identified, other than previously characterised histone acetyltransferases (HATs) and histone deacetylases (HDACs) (Table 2). Although there is some evidence that histone Lys crotonylation can also occur non-enzymatically, this may be an artefact of in vitro conditions in the presence of a high concentration of crotonyl-CoA (Liu S. et al., 2017).

Early recognition of the physiological relevance of histone crotonylation prompted studies into the identification of candidate chromatin-associated proteins that are able to “read” this mark. These efforts focused on classical members of the three major families of histone Kac readers, which were examined for their ability to recognise the unique structure of the crotonyl group conjugated to histone lysine residues. These include bromodomains, YEATS (Yaf9, ENL, AF9, Taf14, and Sas5) and double plant homeodomain finger (DPF) domains proteins (Sabari et al., 2017; Figure 2). By analysing the crystal structure of the human AF9 YEATS domain in complex with H3K9ac, Li et al. (2016) speculated that the YEATS domain could preferentially accommodate longer and bulkier acyl groups, due to an open space within the binding pocket (Figure 2A). This was substantiated in subsequent studies where the YEATS domain was found to have a preference for binding acyl chains longer than acetyl, with the strongest affinity for Kcr (Li et al., 2016). Notably, YEATS domains have a preference for Kcr binding by ∼2–7-fold compared to Kac (Zhao et al., 2017). By using a peptide array and isothermal titration calorimetry it was revealed that AF9 YEATS recognizes histone H3 crotonylation at K9, K18, and K27 with highest affinity for H3K9cr. In addition, the YEATS domain of yeast Taf14 was found to have a similar preference for binding to histone H3 crotonylation as AF9 YEATS. In contrast, the YEATS domain of YEATS2 is selective for histone H3K27cr (Li et al., 2017). The preferential binding of the YEATS domain to sites of Kcr is a result of a unique “aromatic-π-aromatic” stacking (also called “π-π-π” stacking) where the planar crotonylamide group is sandwiched by two aromatic residues, in addition to hydrophobic interactions introduced by a hydrocarbon extension. This aromatic π stacking mechanism for Kcr recognition is consistently observed in the crystal structures of AF9, YEATS2, and Taf14 in complex with Kcr (Figure 2B; Andrews et al., 2016; Li et al., 2016; Zhang et al., 2016; Zhao et al., 2016). In contrast, bromodomains adopt a side-open pocket that generates a spatial restraint that limits their interaction with Kcr. The second bromodomain of TAF1 and the BRD9 bromodomain are among the few examples of this type of reader module that can bind Kcr, albeit with a limited affinity compared to its Kac cognate (Figure 2A; Flynn et al., 2015). Therefore, the YEATS domain proteins represent the first class of selective Kcr readers.

Based on the resolved structure of the YEATS-Kcr complex, Li et al. (2018) developed the first class of YEATS domain selective inhibitors. Among a series of peptide-based molecular probes, one was optimised to selectively target the ENL YEATS domain (XL-13m), which has been previously implicated in the regulation of the oncogenic transcriptional program in acute leukaemia (Figure 2C). The peptide XL-13m was found to associate with endogenous ENL, disrupting the recruitment of ENL onto chromatin, and synergizing with BET and histone methyltransferase DOT1L inhibitors, leading to enhanced downregulation of a set of oncogenes in MLL-rearranged acute leukaemia (Li et al., 2018). Given the preference of YEATS domain proteins for Kcr over Kac, inhibitors developed in this study can be exploited to further investigate the physiological and pathological role of histone Kcr.

The DPF domains of the MYST family member monocytic leukemic zinc-finger (MOZ, also known as KAT6A) and DPF2 also exhibit diverse reader activity with the highest binding affinity for Kcr (Xiong et al., 2016). This was revealed using isothermal titration calorimetry, where DPF domains were found to have a 4–8-fold enhanced affinity for Kcr compared to Kac, while in affinity pull-downs MOZ and DPF2 were found to have specificity for H3K14cr. Although the DPF domains display similar Kcr selectivity as YEATS, the underlying mechanism for this preference in distinct. In the crystal structure of the DPF domain of MOZ in complex with H3K14cr, it was revealed that the interaction was through an intimate hydrophobic pocket that lacked aromatic sandwiching residues, which are characteristic of the YEATS domains (Figure 2A). The importance of this mechanism of Kcr interaction with DPF domains was revealed using ChIP-qPCR and immunofluorescence, where MOZ colocalised with sites of H3K14cr in cells in a manner dependent on its DPF domain (Xiong et al., 2016). In a more recent study, the DPF domain of the HAT complex MOZ-related factor (MORF), was also shown to preferentially bind H3K14cr over H3K14ac. Moreover, binding of the DPF domain to H3K14cr enhanced the catalytic activity of MORF towards another acetylation site (H3K23ac), highlighting the interrelated nature of different PTMs (Klein et al., 2019).

The donor for histone crotonylation, crotonyl-CoA, is an important intermediate involved in several cellular metabolic pathways including fatty acid and amino acid metabolism. The synthesis of crotonyl-CoA can also occur in the mitochondria or the cytoplasm. However, the metabolic sources and mechanisms responsible for generating the nuclear pool of crotonyl-CoA that fuels histone crotonylation remain unknown. Mounting evidence suggests that histone acylations are directly sensitive to changes in the concentrations of their corresponding acyl-CoA metabolites, and therefore can act as indicators of the cellular metabolic state (Simithy et al., 2017).

Sabari et al. (2015) provided initial evidence that histone crotonylation can be regulated metabolically through pathways that influence the cellular concentrations of crotonyl-CoA (Figure 1). Here the addition of the short chain fatty acid (SCFA) crotonate to HeLa S3 cells was found to dramatically increase both the cellular concentration of crotonyl-CoA and H3K18cr in a dose-dependent manner. Crotonate, like other SCFAs, is mainly produced by the gut microbiota during the fermentation of partially and nondigestible carbohydrates (Tan J. et al., 2014). Circulating SCFAs (acetate, crotonate, butyrate, and propionate) can be taken up by tissues and converted into their cognate short-chain acyl-CoAs, the direct donors of histone Lys acylations. This is consistent with depletion of the gut microbiota in mice with antibiotics, which led to a reduction in luminal and serum SCFAs with a concomitant decrease in histone crotonylation (Fellows et al., 2018). Nevertheless, endogenous sources of crotonate are unclear. The mechanism for the action of crotonate in HeLa cells was explained by its cellular uptake, leading to its conversion to crotonyl-CoA by the metabolic enzyme acyl-CoA synthetase 2 (ACSS2 or AceCS1), which is also known to generate acetyl-CoA from acetate. Intriguingly, the knock-down of ACSS2 led to a reduction in basal histone Kcr, suggesting that crotonate might be a physiologically relevant source of crotonyl-CoA. However, the ability of ACSS2 to synthesize crotonyl-CoA from crotonate has not been directly demonstrated with in vitro assays. Hence, the possibility of an indirect reduction of histone crotonylation upon ACSS2 knockdown cannot be ruled out. Indeed, ACSS2 knock-down leads to a reduction of the cellular pool of acetyl-CoA, which is used for fatty acid synthesis and hence beta oxidation of fatty acids, both required for crotonyl-CoA production. In fact, an alternative potential source of crotonyl-CoA in metazoans is through metabolic pathways that include fatty acid β-oxidation or the metabolism of the essential amino acids lysine or tryptophan. The sequential breakdown of fatty acid molecules by mitochondrial β-oxidation to form acetyl-CoA, leads to generation of a crotonyl-CoA intermediate upon oxidation of butyryl-CoA, catalysed by acyl-CoA dehydrogenase. In addition, degradation of the essential amino acids lysine or tryptophan within the mitochondria also generates a crotonyl-CoA intermediate upon oxidative decarboxylation of glutaryl-CoA, catalysed by glutaryl-CoA dehydrogenase. Once crotonyl-CoA is formed by either the fatty acid β-oxidation or amino acid degradation pathways, it undergoes hydration to 3-hydroxybutyryl-CoA that is catalysed by enoyl-CoA hydratase. However, some crotonyl-CoA may escape degradation and instead leak from the mitochondria where it can then contribute to histone crotonylation. This would be consistent with evidence that histone crotonylation is mediated by the fatty acid β-oxidation pathway in yeast (Gowans et al., 2019).

Based on the dual enzymatic activity of p300, it was speculated that crotonyl-CoA and acetyl-CoA could compete to influence the type of acylation. So far very few studies have attempted to measure the relative intracellular concentrations of acetyl-CoA and crotonyl-CoA. In such an effort, crotonyl-CoA was found to be ∼1,000-fold less abundant compared to acetyl-CoA in various cell types (HeLa cells and myogenic cells) (Sabari et al., 2015; Simithy et al., 2017). However, when the cellular pool of citrate-derived acetyl-CoA is depleted by knocking down ATP citrate lyase, this not only decreases acetylation of H3K18 as expected (Wellen et al., 2009), but also increases p300-catalysed crotonylation (Sabari et al., 2015). In addition, when pyruvate dehydrogenase is knocked down, an enzyme recently found to produce a nuclear pool of acetyl-CoA from pyruvate needed for histone acetylation (Sutendra et al., 2014), this also promoted H3K18cr. Together these findings support a model where crotonyl-CoA can compete with acetyl-CoA for p300’s acyltransferase activity.

The repertoire of histone PTMs has greatly expanded in recent years, adding further complexity to the field of chromatin biology (Barnes et al., 2019). This includes, the discovery of a group of short-chain Lys acylations that are structurally similar to Kac and provide a link between cellular metabolism and gene regulation (Sabari et al., 2017). Among these newly identified modifications, histone Kcr is an evolutionary conserved epigenetic mark with a pronounced ability to regulate gene expression (Tan et al., 2011; Sabari et al., 2015). Since its discovery there has been mounting evidence for the functional importance of this modification.

Originally histone Kcr was perceived to fulfil a similar role to Kac, as both share sites of modification, writers and erasers (Zhao et al., 2018). However, the identification of reader modules, such as the YEATS domains proteins, that selectively recognize Kcr indicates that these two modifications are interpreted differentially, thus providing diversity in their functional outcome (Li et al., 2017). In addition, the extent of histone Kcr is also metabolically regulated by the cellular concentration of crotonyl-CoA (Sabari et al., 2015). Therefore, changes in the relative abundance of intracellular concentrations of crotonyl-CoA and acetyl-CoA will influence their respective acylations, mediated by the HAT and co-activator p300. This coupled to difference in reader interaction provides a mechanism for diverse signalling. Furthermore, some genomic regions, such as the “escapee” spermatogenesis genes, are also exclusively marked by Kcr and not Kac (Tan et al., 2011). How histone Kcr can be exclusively marked remains unknown. It is anticipated in future studies, novel proteins that interact with or regulate sites of crotonylation will be identified, including Kcr modifying enzymes or additional Kcr selective readers. This may also include writers or erasers that are specific for histone Kcr, which may explain how some sites can be exclusively marked with this modification.

Despite evidence supporting a unique role of histone Kcr in the regulation of specialised transcriptional programs, one cannot overlook the existing similarities between histone crotonylation and histone acetylation. In fact, ChIP-seq mapping with parallel analysis of histone crotonylation and histone acetylation demonstrated a significant overlap between these two marks at genomic locations in human somatic cells (Tan et al., 2011). This observation also holds true for other recently discovered short-chain histone acylations. For instance, ChIP-seq analyses mapped histone acetylation to similar locations, mostly promoters, with histone propionylation and histone butyrylation in hepatic, and spermatogenic cells (Goudarzi et al., 2016; Kebede et al., 2017). More interestingly, the co-occurrence of these marks at transcription start sites of active genes was positively correlated with gene expression. This suggests that these marks are likely to act in combination to promote a high transcriptional outcome. In spermatogenic cells, a competing nature between histone butyrylation and histone acetylation was reported (Goudarzi et al., 2016). Despite acting as a transcriptional activator, histone butyrylation was found to compete with acetylation to prevent binding of the testis specific BET bromodomain factor, Brdt, an important protein that regulates gene expression and histone-to-protamine transition. In late spermatogenic stages, histone acetylation is important for Brdt-dependent histone removal. Butyrylated histones survive this wave of acetylation-dependent histone removal which is consistent with the inability of Brdt to recognize butyrylation. Although histone crotonylation was not examined in this study, the inability of BET factors to bind histone crotonylation could also explain the identified persistence of histone marks in elongating spermatids (Tan et al., 2011; Flynn et al., 2015). These studies suggest that the rapid turnover between modifications, that in turn allows a dynamic association of reader proteins might be important for diverse transcriptional responses and therefore, these marks are likely to act together to coordinate particular transcriptional programs.

Histone Kcr can regulate diverse physiological functions ranging from gene activation to spermatogenesis (Sabari et al., 2015; Liu S. et al., 2017). Furthermore, it can mediate both protective and adverse functions in the development of different diseases (Ruiz-Andres et al., 2016; Liu et al., 2019). However, only a small number of Kcr sites in human histones have been identified so far (Tan et al., 2011). This is in part due to the lack of commercially available Kcr site-specific antibodies, which has meant much of the research in this field has focused on studying total histone crotonylation. This is likely to limit our understanding of the importance of histone Kcr, as the functional impact of modification at specific sites cannot be readily assessed. Moreover, as the histone code dictates, epigenetic responses are a result of a complex interplay between different PTMs (Bannister and Kouzarides, 2011). Furthermore, as the writers and erasers that regulate histone crotonylation also mediate histone acetylation, it is often difficult to discern the functional role of each modification. Although CDYL has been used to target histone crotonylation, due to its nuclear crotonyl-CoA hydratase activity, it lacks specificity as it also functions as a reader of methylated histone, as well as having other transcriptional repressive activities (Zhang et al., 2011; Liu Y. et al., 2017; Liu et al., 2019). Therefore, there is a need for improved access to tools to study histone Kcr, to allow a greater understanding of how this modification contributes to the regulation of physiological and pathological processes.

The intracellular abundance of crotonyl-CoA is considered relatively low, therefore small fluctuations in crotonyl-CoA concentration will likely have a pronounced impact on crotonyltransferase reactions that mediate histone Kcr (Sabari et al., 2015). Accordingly, the investigation of pathways that fuel the generation of crotonyl-CoA destined for histone crotonylation, will improve our understanding of processes that regulate this modification. Although crotonyl-CoA is an intermediate in metabolic processes including fatty acid β-oxidation and degradation of the essential amino acids lysine or tryptophan, the contribution of these pathways in mediating histone crotonylation in mammalian cells remains unknown. If a link between metabolism and histone crotonylation is established then this modification is likely to have important implications in prevalent diseases associated with metabolic dysfunction, which include cancer and cardiovascular disease.

In summary, histone crotonylation has an active role in gene regulation that is functionally distinct from histone acetylation. The pronounced ability of histone Kcr to regulate gene expression, which outperforms histone Kac, has helped to establish this modification as an important regulator of cellular signalling and tissue function. However, further studies are needed to better define how histone crotonylation is regulated and its association with diverse physiological and pathological processes.

AN wrote the manuscript. JB made suggestions and edited the manuscript. Both authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.