Histone acylation at a glance

Interestingly, histones undergo a variety of functional post-translational modifications. The first evidence of histone modifications was presented in the early 1960s by Allfrey and colleagues; by using a labeled [¹⁴C]acetate precursor, they found that histones in isolated calf thymus nuclei had undergone acetylation (Allfrey et al., 1964). They also observed that acetylated histones lose much of their effectiveness as inhibitors of RNA synthesis, thus revealing the possible function of these marks in regulating transcription.

Indeed, histone acetylation positively correlates with gene activation, the best characterized of which are the acetylation marks on histone H3, including the transcription-activating marks H3K9, K14 and K23 acetylation, as well as H3K27 acetylation, which might mark active enhancers (Strahl and Allis, 2000). Some marks, however, are associated with both transcriptional activation and repression (e.g. methylation), depending on the position of the modified residue on histone tails (Jenuwein and Allis, 2001). Considering the variety and pervasiveness of these histone marks, there has been immense interest in studying histone-modifying enzymes, which has resulted in the identification of proteins that deposit (writers) and remove (erasers) histone modifications, as well as binders (readers) that bring about downstream effects (Bannister and Kouzarides, 2011; Arrowsmith et al., 2012; Yang and Seto, 2008). Interestingly, recent studies have demonstrated that although acetylation is the predominant form of modification, there are numerous other acylations that histones undergo that differ in hydrocarbon chain length, hydrophobicity or charge. Advances in mass spectrometry technology have led to the identification of such histone modifications, including butyrylation (Bu), succinylation (Suc), crotonylation (Cr), propionylation (Pr) and glutarylation (Glu), among others (see poster) (Tan et al., 2011; Xie et al., 2012; Sabari et al., 2015). There is growing interest in understanding their possible roles, and interesting nuances have begun to emerge. For example, histone H4K5 butyrylation competes with acetylation to prevent the binding of the bromodomain-containing protein Brdt (Goudarzi et al., 2016). Moreover, the removal of H4K5K8 butyrylated histones from the chromatin is retarded during late spermatogenesis (Goudarzi et al., 2016). During starvation, the β-hydroxybutyryl (βhb) mark at H3K9 distinguishes a set of upregulated genes from others that bear H3K9ac and H3K4me3 (me3, tri-methylation) marks, suggesting that histone lysine βhb performs different functions from acetylation (Xie et al., 2016). Therefore, even though both acetyl and butyryl groups are stimulators of transcription, their impacts on cell physiology differ. Hence, such findings suggest that all acyl groups are not functionally equivalent. This increases the complexity of the histone code, which proposes that specific combinations of histone modifications impact chromatin–DNA interactions and gene expression (Strahl and Allis, 2000).

In this context, an often-overlooked aspect is the relationship between the metabolic state of cells and the above-mentioned modifications. The cofactors and substrates required for the deposition of histone marks are frequently metabolites, such as acetyl-coenzyme A (acetyl-CoA) for acetylation, S-adenosyl methionine (SAM) for methylation and ATP for phosphorylation. Each acylation modification derives from a specific acyl-CoA metabolite, such as butyryl-CoA, succinyl-CoA, crotonyl-CoA or glutaryl-CoA. In this Cell Science at a Glance article, focusing on these additional modifications, we provide a general overview of the basic principles of how histone acylations impart their function and the regulation of these epigenetic signals.

Among the histone acyl marks, histone acetylation is the best studied. The distribution of histone acetylation, both in the context of genomic features and regarding the position of the lysine residues on histone tails, is governed by histone-modifying enzymes. Studies to date suggest that the enzymes that deposit the non-acetyl acylation marks are the same as those that acetylate histones (Nitsch et al., 2021; Fellows et al., 2018; Huang et al., 2018a). These enzymes are known as histone acetyltransferases [HATs, also known as KATs for lysine (K) acetyltransferases; reviewed by Marmorstein and Zhou, 2014] (see poster). For example, one such well-known HAT, GCN5 (also known as KAT2A), has been reported to perform histone succinylation at the H3K79 position (Wang et al., 2017).

Besides enzymatic deposition, non-enzymatic protein acylation has been reported (Trub and Hirschey, 2018; Cai and Tu, 2011). Unlike acetyl-CoA, succinyl-CoA and glutaryl-CoA can chemically acylate proteins on lysine residues efficiently (Wagner et al., 2017). This happens through the nucleophilic attack on the carbonyl carbon of acyl-CoAs by the unprotonated ε-amino group of lysine residues. This reaction mechanism is facilitated by an alkaline pH and higher concentrations of acyl-CoA molecules (Paik et al., 1970; Baeza et al., 2015). Although non-enzymatic acylation can be especially relevant in the mitochondrial matrix, which exhibits a more alkaline pH (Wagner and Hirschey, 2014; Kulkarni et al., 2017; Wagner et al., 2017), the inherently higher electrophilicity of succinyl-CoA and glutaryl-CoA might promote their reaction with histone tails non-enzymatically. Indeed, in vitro studies have revealed that the protein undergo non-enzymatic acylation (Maksimovic and David, 2021). Interestingly, the most prevalent sites observed for non-enzymatic acylations on histones are closer to the C-terminus of the protein, whereas enzymes show higher specificity for sites at the N-terminal tails (Simithy et al., 2017). Hence, histone acylations in cells could be the outcome of both enzymatic and non-enzymatic mechanisms. However, when evaluating the evidence for non-enzymatic covalent modification (NECM) of proteins, an important factor to consider is the concentration of the co-factor used in the in vitro assay. The concentration of acyl-CoAs in the cell is in the micromolar range; however, the concentration used in in vitro assays is often ∼10 times higher than that. It is unclear whether the local intracellular concentration of acyl-CoA near histones can reach sufficiently high levels to cause NECM.

The enzymes that remove the acyl marks from proteins are known as lysine deacetylases. The 18 lysine deacetylases that have been identified in the human and mouse genomes belong to two distinct families with different catalytic mechanisms: the NAD⁺-dependent sirtuin deacetylases (SIRT1–SIRT7) and the Zn²⁺-dependent histone deacetylases (HDAC1–HDAC11) (Seto and Yoshida, 2014).

Sirtuins are localized to and deacetylate in different compartments within the cell (reviewed by Wang and Lin, 2021). SIRT1, SIRT6 and SIRT7 are nuclear proteins, and SIRT1 deletion in mouse embryonic fibroblasts (MEFs) affects nearly 10% of all acetylation sites, most of which occur on nuclear proteins (Chen et al., 2012). By contrast, SIRT3, SIRT4 and SIRT5 are mitochondrial, where they are thought to be especially important for the removal of non-enzymatic acylations, and SIRT2 is a cytoplasmic protein that can translocate into the nucleus (Michishita et al., 2005). A landmark study in 2011 demonstrated that certain sirtuin enzymes, such as SIRT5, preferentially remove other acylation marks (e.g. succinylation and malonylation) as opposed to acetylation (Du et al., 2011). SIRT7 has been reported to remove lysine succinylation marks on histone H3K122 and thereby promote chromatin condensation and DNA double-strand break repair (Li et al., 2016a). Thus, the localization and substrate specificity of sirtuins must be considered to elucidate their roles in the removal of acylations from histones and other protein substrates.

In parallel, the role of HDACs in histone deacetylation is well known and their role in the removal of non-acetyl acyl marks is beginning to emerge. One study found that class I HDACs, rather than SIRTs, are the major histone decrotonylases (Wei et al., 2017). Notably, the authors successfully generated mutants of HDAC1 and HDAC3 that impaired histone deacetylase but retained decrotonylase activity (Wei et al., 2017). A separate study found that a purified ternary complex of HDAC1–CoREST1–LSD1 (CoREST1 is also known as RCOR1, and LSD1 as KDM1A) was able to directly hydrolyze H3K18cr on peptide substrates (Kelly et al., 2018). Moreover, the global levels of histone crotonylation were increased upon the deletion of HDAC1 and HDAC2 in embryonic stem cells. Interestingly, this increase in histone crotonylation largely overlapped with H3K18ac at the transcription start site (TSS) and correlated with increased gene activity (Kelly et al., 2018). Furthermore, in vitro experiments have revealed that HDAC2 and HDAC3 exhibit de-2-hydroxyisobutyrylation activity (Huang et al., 2018a). However, whether they perform a similar function in vivo remains to be demonstrated. Moreover, it is unclear whether other HDACs might exhibit non-deacetyl deacylase activity.

Readers of histone acylations encompass a diverse array of proteins equipped with specialized domains adept at recognizing and binding to acylated histone residues. Readers harboring a Yaf9, ENL, AF9, Taf14 and Sas5 (YEATS) domain, a bromodomain (BRD) or a double PHD finger (DPF) domain are crucial in recognizing and interpreting acylations within the context of chromatin (see poster) (Marmorstein and Zhou, 2014). Each of these reader domains exhibits distinct structural features and binding specificities, allowing them to orchestrate diverse chromatin-related processes (Yun et al., 2011).

BRDs are perhaps the most well-characterized reader domains, known for their ability to recognize acetyl-lysine modifications on histone tails. BRD-containing proteins, such as bromodomain and extra-terminal (BET) proteins, exert profound effects on gene expression by recruiting transcriptional machinery to acetylated histones, thereby promoting transcriptional activation (Sanchez and Zhou, 2009). YEATS domains largely exhibit a unique affinity for non-acetyl acyl modifications, such as crotonyl and succinyl groups (Schulze et al., 2009); this is due to an open space within its binding pocket, with the strongest affinity for crotonylated lysine residues (Li et al., 2016b). DPFs are characterized by the presence of two consecutive PHD finger motifs that can recognize both histone methylation and acylation marks, allowing for versatile regulation of chromatin structure and function (Soshnikova et al., 2020). Overall, YEATS, BRD and DPF reader domains play crucial roles in deciphering the histone code and translating epigenetic modifications into functional outcomes within the cell.

Although histones undergo a variety of acylations, the non-acetyl acyl marks are found on the same lysine residues where acetylation occurs and are seemingly regulated by the same histone-modifying enzymes that govern acetylation. Then what determines the type of acylation at a given site?

Although acyl-selective writers have yet to be discovered, a difference in the preference towards acyl-chain substrates has been observed. For example, the HATs GCN5 and p300 (also known as EP300) exhibit a preference for one acyl group over the others depending on their chain length. Kinetic analysis of the HATs p300 and GCN5 activities against lysine residues with Ac, Pr, Bu and Cr marks revealed progressively slower rates of enzymatic activity with longer chain acyl-CoA substrates (Kaczmarska et al., 2017). This occurs because although the acetyl group is perfectly positioned for catalysis, longer acyl chains are not as efficient at fitting into the active site of HATs (Ringel and Wolberger, 2016; Kaczmarska et al., 2017).

Another key determinant of the deposition of acylation marks is the levels of metabolites, which can vary due to numerous factors, including nutrient availability. For example, cells grown in glucose-rich environments can synthesize pyruvate through glycolysis. Pyruvate can be converted into acetyl-CoA within the mitochondria through pyruvate dehydrogenase or in the cytoplasm. The resulting acetyl-CoA can be used to acetylate histones, specifically H3K9ac, which activates gene expression at the promoters of growth-promoting genes (see poster). When there is a lack of glucose, cells change the balance of metabolites to survive; they take up fatty acids and perform fatty acid oxidation to generate metabolites like βhb. These metabolites can be conjugated to CoA, resulting in the deposition of marks such as H3K9βhb that help the cell adapt to the new environment (Xie et al., 2016) (see poster).

The dependency on metabolite concentration for the deposition of histone acyl marks arises because the dissociation constant (K_D) of many HATs towards acyl-CoA is relatively high. Notably, enzymes have evolved in a manner to match the acyl-CoA concentration in the respective organism. For instance, GCN5 has a K_D for acetyl-CoA of 0.56 µM and its yeast counterparts have one of 8.5 µM (Langer et al., 2002). These numbers fall in the concentration range of acetyl-CoA found in human and yeast cells (Cai et al., 2011; Lee et al., 2014). In fact, the levels of acyl lysine post-translational modifications show a strong positive correlation with the metabolic levels of acyl donors (Simithy et al., 2017). For instance, the concentration of acetyl-CoA is ∼12 times higher than that of propionyl-CoA in HeLa cells. This difference is reflected in histone acylation, with acetylated histone peptides being ∼10 times more abundant than propionylated peptides (see poster) (Simithy et al., 2017).

Interestingly, cellular metabolites have also been found to regulate HDAC activity. For example, βhb inhibits the activities of HDAC1, HDAC2 and HDAC3 (Shimazu et al., 2013), and the metabolites L-carnitine and sphingosine-1-phosphate have been shown to inhibit HDAC1 and HDAC2 (Huang et al., 2012; Hait et al., 2009). Pyruvate also inhibits the activity of HDAC1 and HDAC3 (Thangaraju et al., 2006). Therefore, in addition to serving as precursors for cofactors, metabolites can directly bind to histone modifiers to regulate epigenetic changes.

It is important to consider the relative abundance of newly discovered histone acylations when trying to understand their function and implications. Advancements in mass spectrometry methods now allow quantification of histone modification abundance, revealing valuable information. In HeLa cells, sites such as H3K9 and H3K14 show similar levels of propionylation and acetylation marks. However, at H4K16, only 0.03% of peptides have the propionyl mark, whereas ∼18% have the acetylation mark (Simithy et al., 2017). In fact, most non-acetyl acylation modifications exhibit extremely low stoichiometry, which might call into question their physiological significance. Acetylation is the most prevalent lysine acylation type among all core histones, making up 67% of the total (see poster) (Simithy et al., 2017). Other non-acetyl acylations, such as succinylation, propionylation and crotonylation, account for the remainder.

Although some of these numbers might appear low, it is important to note that histones are among the most abundant proteins in cells. Hence, overemphasizing the lack of abundance of their modifications might underestimate the potential significance of such marks. This is exemplified by the H3K9ac mark, which is found on only 0.4% and 0.7% of total histone peptides in HeLa cells and myotubes, respectively (Simithy et al., 2017). It is well-documented that H3K9ac marks promoters of actively transcribed genes, despite their seemingly low overall stoichiometry (Kouzarides, 2007). Furthermore, ChIP-seq experiments for lysine residues marked with Cr, Bu, Hib and βhb marks illustrate that these histone acylations also appear to mark active regulatory elements (Goudarzi et al., 2016; Dai et al., 2014; Xie et al., 2016). Therefore, the association of histone acylations with genomic features might provide more meaningful clues about their functional significance.

In addition to the action of writers and erasers, the enzymes that generate acyl-CoA play a crucial role in regulating cellular acylation. In yeast, acetyl-CoA is produced by acetyl-CoA synthetases 1 and 2 (Acs1 and Acs2) (Van den Berg and Steensma, 1995). In vertebrates, ATP citrate lyase (ACLY) is considered the major producer of acetyl-CoA (Chypre et al., 2012).

It is notable that acyl-CoA producing enzymes are found in the nucleus. For example, the Acs2 enzyme (Acss2 in vertebrates), which converts acetate into acetyl-CoA, is found in the nuclei of yeast, mouse and human cells. Moreover, the proportion of Acss2 in the cytoplasm and nucleus varies depending on the metabolic state in both mice and humans (Huang et al., 2018b; Li et al., 2017; Bulusu et al., 2017). This indicates that the localization of Acss2 is influenced by the metabolic needs of the cell in different subcellular compartments. Besides Acs2, both ACLY and the pyruvate dehydrogenase complex (PDC) have also been reported to be present in the nucleus of mammalian cells (Sivanand et al., 2017; Sutendra et al., 2014). PDC is believed to play a role in the zygotic genomic activation of the embryo (Nagaraj et al., 2017), the process during embryonic development where the zygote starts to use its own DNA to control development, rather than relying on maternal RNA and proteins. The knockdown of nuclear PDC in isolated functional nuclei decreased the de novo synthesis of acetyl-CoA and acetylation of core histones (Sutendra et al., 2014). These observations are surprising, especially given that PDC is a megadalton-sized, mitochondrial multiprotein complex. It remains a mystery how the complex is imported into the nucleus and subsequently assembled correctly for its activity. Another notable example is the nuclear presence of the α-ketoglutarate dehydrogenase (α-KGDH) complex in human cell lines where it reportedly binds to KAT2A to mediate succinylation of histones at the promoter regions of genes (Wang et al., 2017).

There might be several reasons why metabolic enzymes are needed in the nucleus. Acyl-CoAs are unable to traverse organellar membranes on their own, such as those of mitochondria. This limitation is thought to constrain the impact of fluctuations in acyl-CoA levels resulting from metabolic changes on the gene regulatory acylation that takes place in the nucleus (Choudhary et al., 2014). Moreover, owing to the higher dissociation constant (K_D) of HATs, their activity can be susceptible to even modest fluctuations in the acyl-CoA concentration. Besides, different acylations are differentially enriched at genomic locations in a context-dependent manner, such as the metabolic state of cells or disease (Sabari et al., 2017; Trefely et al., 2020). These considerations suggest that for histone acylation, cells might likely need to produce acyl-CoA locally in the nucleus. The discovery of metabolic enzymes in the nucleus, and their movement based on metabolic state, suggests a link between the nuclear production of acyl-CoA metabolite and corresponding histone acylation. Nonetheless, future research to understand whether nuclear metabolic enzymes are indeed necessary for histone modification regulation is pertinent.

Metabolites are essential for protein acylation, which means that metabolic disorders can have an impact on the protein acylome. Acylcarnitines are formed by the combination of acyl groups with carnitine and help transport acyl groups across membranes. Increased levels of acyl-CoA can result in the increased accumulation of the corresponding acylcarnitine. For example, genetic defects can result in the inactivation of malonyl-CoA decarboxylase (MLYCD), a human enzyme that catabolizes malonyl-CoA. This results in the marked accumulation of malonyl-CoA and malonylcarnitine (Colak et al., 2015; Pougovkina et al., 2014). Several studies have found increased levels of acyl-CoAs and/or acylcarnitines in tissue and plasma in metabolic disorders, such as obesity, diabetes, heart failure and inborn errors of metabolism (McCoin et al., 2015; Newgard, 2017). Metabolic disorders are also influenced by acyl writers, readers and erasers; for example, mice lacking the mitochondrial deacetylase SIRT3 predictably exhibit mitochondrial protein hyperacetylation, among other metabolic phenotypes (Hirschey et al., 2011; Hebert et al., 2013).

A different scenario is presented in cancers, where a global decrease in histone acetylation is often observed (Glozak and Seto, 2007). In contrast to normal cells that primarily rely on glucose for growth, cancer cells can exhibit acetate dependency and often have higher expression of HDACs (Comerford et al., 2014; Schug et al., 2015). To target this characteristic, HDAC inhibitors such as suberoylanilide hydroxamic acid (SAHA), have been approved as cancer therapeutics (Mann et al., 2007; Ropero and Esteller, 2007). Another method for counteracting aberrant lysine acylations is through sirtuins. SIRT1 activators are being tested clinically and pre-clinically in a wide variety of diseases (Sinclair and Guarente, 2014). Increasing NAD⁺ levels by providing cells or mice with NAD⁺ precursors, such as nicotinamide riboside (NR) and nicotinamide mononucleotide (NMN), has been shown to activate SIRT3 (Cantó et al., 2012; Karamanlidis et al., 2013). These NAD⁺ precursors, as well as specific sirtuin activators, might be useful for targeting aberrant lysine acylation in metabolic disorders.

Notably, sirtuins are known to regulate the process of aging by suppressing cellular senescence (Zhao et al., 2020). This process is believed to occur through the delay of age-related telomere attrition (shortening of telomeres with age), maintenance of genome integrity and promotion of DNA damage repair (Morris, 2021). The deacetylase activity of SIRT1, for example, which deacetylates histones and over 50 non-histone proteins, including transcription factors and DNA repair proteins (Michan and Sinclair, 2007), is a crucial factor in regulating senescence. The reversible nature of acylations makes them a promising target for therapeutic interventions and continued efforts in this area are likely to result in drug targets.

The increasing number of reported histone acylation modifications suggests an increasing complexity of the histone code and a myriad of possible mechanisms to regulate gene expression in tune with metabolic cues, as reflected by the availability of particular acyl-CoA substrates. However, more work along with rigorous experimentation is needed to establish their potential physiological significance, especially given that there are still concerns about the specificity of antibodies used in investigating these marks (Tsusaka et al., 2023). Important considerations, such as the stoichiometry of such modifications, the source and subcellular compartmentalization of the acyl-CoA substrates, and the activity of dedicated writers and erasers, must also be considered. The existence of specific erasers of acylation (i.e. sirtuins) suggests that these modifications do indeed occur inside cells; however, to date, evidence of specific writers or readers of acylations beyond acetylation remains sparse, consistent with the hypothesis that many of these acylations are non-enzymatic and spurious, thereby necessitating their removal to preserve protein functions.

The authors are grateful to the members of the Tu lab for their useful suggestions.