Histone H2A Variants: Diversifying chromatin to ensure genome integrity

Abstract

Histone variants represent chromatin components that diversify the structure and function of the genome. The variants of H2A, primarily H2A.X, H2A.Z and macroH2A, are well-established participants in DNA damage response (DDR) pathways, which function to protect the integrity of the genome. Through their deposition, post-translational modifications and unique protein interaction networks, these variants guard DNA from endogenous threats including replication stress and genome fragility as well as from DNA lesions inflicted by exogenous sources. A growing body of work is now providing a clearer picture on the involvement and mechanistic basis of H2A variant contribution to genome integrity. Beyond their well-documented involvement in gene regulation, we review here how histone H2A variants promote genome stability and how alterations in these pathways contribute to human diseases including cancer.

1. Introduction

1.1. Chromatin and histone variants

Genetic information stored in nuclear DNA of eukaryotes must be compacted to fit into the volume of the nucleus while also being accessible to DNA-templated processes including transcription, replication and DNA repair. This is accomplished by packaging and organizing DNA into chromatin, the basic unit being the nucleosome. Nucleosomes consist of an octamer of histone proteins, which are assembled into higher order structures through linker histones, RNA, and various chromatin interacting proteins. In addition to the core histones H2A, H2B, H3 and H4 that make up the majority of nucleosomes in mammals, variants of core histones also exist that can replace their core histone counterparts and create nucleosomes with altered structural and functional capabilities. In humans, variants of all core histones have been identified, with H2A variants being the most diverse and numerous (Fig. 1a). Of note, there are many copies of each core histone including 17 individual histone genes alone that encode H2A. Surprisingly, even within these 17 genes there is sequence variation of up to several amino acids [1]. H2A histone variants consist of 11 genes that, unlike core histones, can contain introns (ex. H2A.Z.2 and macroH2A.1; Fig. 1b) and encode for at least 13 unique proteins due to alternative splicing. Differences in sequence identity among these variants also range broadly compared to H2A, from only a few amino acids to as little as ~18-% identity. Core histones are expressed mainly in S-phase while expression of histone variants occurs throughout the cell cycle, making their deposition and regulation independent of DNA replication. Analysis of the expression of these variants has revealed tissue and cellular heterogeneity, which also has been shown to be altered in some diseases including cancer [2].

Fig. 1.

Human histone variants. (a) Complete list of human histone variants. Histone H2A variants consist of the highest number of variants (11) while histone H4 has only one variant reported to date in human cells. (b) Human histone H2A variants. Comparison of histone H2A variants with core H2A genes. Number of genes, gene names, protein size in amino acids and percent identity of variants to core H2A, along with key features of each variant, are provided. Note that H2A.Z.2.1 and H2A.Z.2.2, as well as macroH2A.1.1 and macroH2A.1.2, are splice variants of the same gene.

1.2. Post-translational modifications

The presence of numerous histone variants allows for an incredible potential for nucleosome diversity that can configure and fine-tune chromatin structure and function. Post-translational modifications (PTMs) of histone variants provide yet another level of complexity and diversity (Fig. 2). While writer enzymes have been defined for several of these PTMs (see Fig. 2), many remain poorly characterized and their biological function and regulatory mechanisms undefined. PTMs including phosphorylation, acetylation, methylation, ubiquitylation, SUMOylation and PARylation have been shown to decorate histone variants, which can provide binding sites for chromatin reader proteins that allow these factors to exert their function in a locally and often temporally controlled manner [2, 3]. This is equally true for the binding of DNA repair factors to PTM-mediated signals surrounding damaged DNA, which significantly impact chromatin and DNA repair processes [3, 4].

Fig. 2.

Post-translational modifications of histone H2A variants. Key PTMs of mammalian histone H2A variants with their modified residue (if known) are shown. Examples of histone writer enzymes for specific PTMs are included. Asterisks indicate variant specific modifications among closely related variants. Please note the list of histone H2A variant PTMs and their writer enzymes is not comprehensive.

1.3. Histone H2A variants in genome integrity and disease

While histone variants play well-established roles in regulating chromatin during transcription and development, which have been comprehensively reviewed elsewhere [2], emerging evidence reveals their important contribution for maintaining genome integrity. Here we focus our discussion on the involvement of mammalian histone H2A variants and their PTMs in DNA damage response pathways that respond to endogenous and exogenous sources of DNA damage, including during DNA double-strand break (DSB) repair (Fig. 2). Given that genome instability is a hallmark or enabling defect of several human diseases such as cancer, neurodegeneration and premature aging [5], we will further discuss how recent disease-related H2A variant biology may relate to their function in genome maintenance.

The complexity and necessity of maintaining the integrity of the genome is perhaps best characterized in the context of DSB repair, which represent dangerous lesions that if left unrepaired, can result in mutations and cell death. The DNA damage response (DDR) counters DSBs through a series of tightly controlled, sequential events that initiate with DNA end sensing, poly(ADP-ribose) polymerase (PARP) activation and concomitant chromatin relaxation, which is followed by the recruitment of the DDR kinases ATM, ATR and / or DNA-PKcs as well as a number of chromatin modifiers and remodelers that together recruit and activate downstream repair effector proteins [6–8]. The precise nature of these factors depends on the type of DSB repair, which in turn depends on the cell cycle phase as well as the location of the DNA lesion and has been summarized in excellent detail in several recent reviews [9–13]. In brief, there are two main DSB repair pathways: nonhomologous end-joining (NHEJ) which mediates the ligation of broken ends in a template-independent manner and homologous recombination (HR), which facilitates error-free repair using a homologous, undamaged DNA template [10, 12, 14–16]. Alternative end joining (Alt-EJ), which includes Polymerase theta mediated end-joining (TMEJ), represents a third, less frequently used group of DSB repair pathways that are normally suppressed by HR or NHEJ [12, 17]. Alt-EJ repairs DNA ends using small 3 - 25 bp patches of homology and is thus also error prone by nature, and often referred to as microhomology-mediated end joining (MMEJ) [10]. Both HR and Alt-EJ depend on DNA end resection, a process involving various endo- and exonuclease activities, such as MRE11 of the DNA end sensing MRE11-NBS1-RAD50 (MRN) complex, the 5’ to 3’ exonuclease EXO1, and others. In the case of HR, long-range resection is further supported by CtIP and the BRCA tumor suppressor genes [10, 17]. Resected single-stranded DNA (ssDNA) is coated with the ssDNA binding protein RPA, which is replaced by the RAD51 recombinase that forms filaments involved in homology search and strand invasion. The activity of BRCA1 and subsequent HR initiation is counteracted by 53BP1 in a manner that is tightly controlled by the underlying chromatin environment and thereby modulates the choice of repair pathway that acts on each individual DSB (see below and [18–21]).

Defects in DSB repair pathways and many of the underlying repair factors result in genome instability that can involve point mutations, small insertions and deletions, as well as larger chromosomal aberrations and translocations, all of which have been associated with human pathologies [5]. Repair pathway imbalance can further result in pathway-specific vulnerabilities, which has been exploited as a powerful strategy to develop new targeted therapies to treat cancer, a concept first used to develop PARP inhibitors (PARPi) to target HR-deficient tumors [22, 23]. In the following, we will discuss how H2A variants participate in DSB repair through their ability to alter the DNA lesion-surrounding chromatin context in a way that directly impacts chromatin accessibility, sensing of the DNA lesion, repair factor recruitment and, ultimately, optimal DNA repair. As defects in H2A variant function have been associated with a number of human pathologies, we further discuss the potential involvement of their genome repair activities in these diseases, which may be distinct from or complementary to their well-known functions in gene regulation and development.

2. H2A.X

2.1. γH2A.X in genome maintenance

The histone H2A variant H2A.X, which comprises ~ 10% of chromatin-associated H2A species, was first described in 1980 by West and Bonner, who used two dimensional histone gel electrophoresis to separate and identify histone H2A variants [24]. Several years later, the same group realized the involvement of H2A.X in the DDR, when they determined that H2A.X from mammalian cells and mice was phosphorylated on serine 139 within the SQEY motif comprising the 4 most C-terminal amino acids of this variant. This modified protein species was termed “γH2A.X” from its discovery following DNA damage induction by ionizing gamma-radiation (IR) [25]. Interestingly, the SQ motif located two amino acids from the end of the histone protein is invariant across species, being found in animals, plants, fungi and even protists [26]. For example, in fission and budding yeast, the core histone H2A is a homolog of H2A.X and contains the SQ motif. Using a specific antibody raised against the S139-phosphorylated H2A.X peptide, immunofluorescence of irradiated mammalian cells and other species revealed discrete nuclear foci of γH2A.X representing megabases of modified chromatin that marked individual DSBs (Fig. 3a) [21]. The ability of γH2A.X to identify unrepaired DNA lesions in cells was a landmark finding that remains the standard technique used by most researchers to label DNA lesions in cells, which is reflected in the number of citations on H2A.X and γH2A.X that continue to be published to this day (Fig. 3b).

Fig. 3.

Characterization of γH2A.X and histone H2A variant publications. (a) Ionizing radiation-induced γH2A.X foci. Human U2OS cancer cells treated with IR and analyzed by immunofluorescence with a γH2A.X specific antibody (image provided by Doohyung Lee, Miller lab, UT Austin). Each focus represents γH2AX signal surrounding an individual DNA double-strand break. γH2AX-modified nucleosomes are highlighted in red and unmodified nucleosomes are in gray. (b) Analysis of publications involving histone H2A variants. H2A.X is the most highly studied histone H2A variants, with the majority of publications involving γH2A.X.

After its initial discovery, γH2A.X was shown to co-localize with several DNA repair proteins including RAD50 of the MRN complex, BRCA1 and RAD51 [28]. Several other groups rapidly identified additional factors that localized to γH2A.X-marked DSBs including ATM, 53BP1 and MDC1 (reviewed in [29]). Using various phospho-H2A.X peptide probes to mimic γH2A.X, MDC1 and 53BP1 were shown to directly bind to H2A.X phosphorylated on S139, identifying these factors as “readers” of γH2A.X within chromatin-surrounding DNA break sites [30–33]. The identification of the kinase(s) responsible for γH2A.X was first hinted at by the observation that treatment with the PI-3 kinase inhibitor wortmannin prevented damage-induced γH2A.X [28]. The ataxia-telangiectasia mutated (ATM) kinase, a phosphoinositide 3 kinase-related protein kinase (PI3KK), was shown to directly phosphorylate H2A.X on S139 within its preferred SQ/TQ motif in vitro, and cells deficient for ATM resulted in defective γH2A.X formation [34]. The related PI3KKs DNA-PK and ATR are also capable of phosphorylating γH2A.X [35, 36]. Upon DDR activation, MDC1 becomes a platform for the recruitment of additional proteins to break sites through both phospho-specific and other interactions (reviewed in [6]). These include NBS1 of the MRN complex [37–40], ATM [41], and the E3 ubiquitin ligase RNF8 [42–44]. The ability of MDC1 to bind both γH2A.X and MRN/ATM provides a mechanism whereby this signal can spread on chromatin across several 100 kb to create the visible foci observed by immunofluorescence in cells. The γH2A.X-MDC1 axis plays a pivotal role in blocking CtIP-mediated DNA end resection in G1 phase lymphocytes to promote V(D)J recombination and genome stability [45]. Thus, γH2A.X serves as a molecular amplifier and platform that acts to recruit and retain DDR factors to DSB-surrounding chromatin to promote their signaling, processing and repair.

The physiological relevance of γH2A.X was first established by the ability to eliminate H2A.X in mice and mammalian cells. The Nussenzweig and Alt labs generated H2A.X null mice, which helped define the physiological role of this variant in suppressing genome instability [46, 47]. Indeed, although H2A.X loss was not essential for viability, H2A.X knockout (KO) mice were born ~30% smaller than their WT littermates and displayed sensitivity to IR. Embryonic fibroblasts derived from H2A.X KO mice grew poorly in culture and displayed premature senescence, consistent with the observation that these cells contained increased levels of endogenous DNA damage. Importantly, the DNA damage-induced foci of repair proteins including 53BP1 and BRCA1 that colocalized with γH2A.X were impaired in H2A.X KO cells as well as in cells complemented with H2A.X containing a non-phosphorylatable S139A mutation, demonstrating that the phosphorylation of H2A.X was specifically required to mount the chromatin response to DSBs. Similar results were observed in mouse embryonic stem cells and human cells lacking H2A.X [48, 49], suggesting a conserved function for H2A.X in DNA repair across mammalian species and cell types.

2.2. PTMs beyond mammalian γH2A.X

As described above and highlighted in Fig. 2, H2A.X is decorated by other PTMs in addition to phosphorylation on S139 and these play additional roles in DNA damage signaling and repair. Once RNF8 is recruited to damage sites, in addition to H2A and H2A.X, it ubiquitinates linker histone H1 with K63-linked ubiquitin chains that are recognized by an additional E3 ligase RNF168, which is required for 53BP1 and BRCA1 recruitment to DNA lesions [50]. RNF168 mono-ubiquitylates both H2A and H2A.X on K15 [51], a mark read by the ubiquitin-dependent recruitment (UDR) domain of 53BP1 [52]. More recently, several groups reported that the obligate E3 ligase partner of the heterodimer complex containing BRCA1 contained a ubiquitin binding domain that also reads K15ub within H2A and possibly H2A.X [53–55]. These studies suggest an elegant competition mechanism between BRCA1 and 53BP1 that regulates DSB repair pathway choice.

Additional PTMs on H2A.X that fine-tune the chromatin response to DSBs have been recently reviewed [3, 20]. For example, H2A.X is sumoylated by PIAS4 (Fig. 2; [49]), which along with another SUMO E3 ligase PIAS1, regulate 53BP1 and BRCA1 dynamics at break sites [56, 57]. The terminal tyrosine of H2A.X is phosphorylated by the non-canonical, bromodomain protein WSTF and dephosphorylated by EYA1 [58, 59]. This mark has been proposed to collaborate with γH2A.X to regulate an apoptotic DNA repair pathway choice, including through a mechanism involving the binding of these marks by the Fe65-JNK1 complex [58]. This mark is also read by the SH2 domain of GRB2, which helps enforce the recruitment of MRE11 to break sites and subsequent HR repair [60]. H2AX has several additional phosphorylations, including on T101, which is induced by DNA damage and contributes to radiation resistance [61]. H2AX is acetylated on K5 and K36 by TIP60 and CBP/p300 respectively, which aids in promoting IR-resistance and HR repair [61, 62]. Methylation of H2A.X on K134 by SUV39H2 promotes γH2A.X formation and DSB repair [63]. PRC1-mediated ubiquitination on K119 is conserved between H2A and H2A.X and has been implicated in DSB repair [64, 65], Poly (ADP-ribose) modification (PARylation) of E141 on H2A.X promotes base excision repair (BER) through the recruitment of NEIL3 glycosylase to oxidative DNA lesions [66]. NEIL3 was shown to bind directly to PAR chains, including ADP-ribosylated H2AX, through its C-terminal GRF zinc-finger domains. Remarkably, E141 PARylation appears to prevent aberrant phosphorylation of nearby S139 (γH2A.X to counteract an erroneous accumulation of DSB repair factors at single-stranded DNA lesions, supporting a regulatory role for H2A.X beyond DSB repair [66]. Together, these studies highlight the importance and complex nature of modifications on H2A.X that are associated with the DDR. However, we still know little about the coordinated regulation of these marks in a spatiotemporal manner and how this relates mechanistically to DNA damage signaling and repair events that are established by γH2A.X Despite its discovery over 20 years ago, future studies are clearly needed to fully comprehend how γH2A.X and its other modifications orchestrate genome maintenance during the DDR and in other biological contexts. Of note, many of the modifications discussed above may not be restricted to H2A.X or H2A. Whether and how other H2A variants are modified, and how such modifications may affect the DDR, remains to be determined.

ATM-mediated phosphorylation of H2A.X is conserved in plants, which express an additional histone H2A variant H2A.W that is also phosphorylated on an SQ residue by ATM [67]. While H2A.X appears to function in euchromatin in plants, H2A.W is localized to heterochromatin where it facilitates genome maintenance pathways including DSB repair, transposon silencing and DNA methylation [67–69]. These studies highlight that while general genome maintenance pathways involving histone H2A variants are conserved from mammals to plants, additional diversification of these pathways has occurred. We refer readers to several excellent reviews focusing on histone variants in plants for additional details [70, 71].

2.3. Anatomy and function of γH2A.X domains

Although the above studies established the functional importance of γH2A.X in DNA damage signaling and repair, the physical nature of γH2A.X damage-induced foci has only recently been investigated. Early work used various γH2A.X chromatin immunoprecipitation (ChIP) techniques to determine that this signal emanated from break sites at senescent telomeres, as well as RAG- and restriction enzyme-induced, site-specific DSBs [72–74]. Of note, these studies observed a non-uniform and asymmetric distribution of γH2A.X spreading from breaks. Using super resolution microscopy, γH2A.X foci induced by IR were further resolved to identify multiple “nano-foci” of γH2A.X associated with each individual DSB with an approximate size of 75 kb, a structure observed as a single focus by traditional microscopy [75]. Notably, these clustered γH2A.X foci were shown to be in close proximity to CTCF, a chromosomal architectural protein involved in organizing topologically associated domains (TADs) and chromosome loops. Similar results were obtained in a study using genome-wide Hi-C chromatin contact maps to map γH2AX propagation as it relates to the spatial organization of TADs and other genomic interactions [76]. Depletion of CTCF resulted in radio-sensitization and a diminution of γH2A.X nano-foci, which strongly suggested that chromatin architecture dictates γH2A.X formation and its function in chromatin-based responses to DNA damage [75]. Related to these findings, super resolution microscopy approaches allowed for the detection of 53BP1 and its effector RIF1 at break-associated topological domains [77]. Cells deficient for 53BP1 or RIF, but not core DSB factors, were unable to maintain these 3-dimensional chromatin domains at DSBs. Finally, using human DIvA (DSB inducible via AsiSI) cells coupled with 4C-Seq and Hi-C, the Legube lab demonstrated that TADs establish γH2A.X-53BP1 domains surrounding DSBs through a loop extrusion mechanism involving CTCF-anchored ATM and cohesins [78]. Thus, the 3D genome organization in cells regulates the establishment of γH2A.X revealing a vital interplay between nuclear architecture and genome integrity. These findings also suggest that variations in genome organization across individual cells, cell types or pathological states may contribute to the engagement, use and efficiency of DNA repair mechanisms that impact genome maintenance [11].

In addition to being modified upon DNA damage, H2A.X has also been shown to be incorporated de novo at sites of UVC-induced and replication damage in a mechanism dependent on the histone chaperone FACT [79]. Conversely, H2A.Z was observed to be removed from break sites, consistent with other studies (see below). This work highlights a potential role for H2A.X deposition directly at break sites in supporting DDR factor recruitment and repair functions at damage sites, beyond its initial phosphorylation. It is worth noting that H2A.X has been reported to promote replication fork stability, and H2A.X deficiency results in sensitivity to replication stress [80, 81]. This process may further involve H2A variant exchanges at stalled forks, given the γH2A.X-dependent deposition of the replication-stress protective macroH2A1 variant at stalled replication forks (see below and [82]). Therefore, the addition and removal of various histone variants likely acts to shape a repair-competent chromatin environment for both the repair of DNA lesions and the epigenetic maintenance of the surrounding chromatin. It is important to note that the initial recruitment of DNA damage response factors is independent of H2A.X [83]. Consistent with this notion, H2AX-independent repair pathways have also been identified. These can involve MDC1 damage recruitment through its interactions with the nucleosome acidic patch, an interaction hub known for coordinating epigenetic and DDR factor chromatin interactions, which promotes 53BP1 damage recognition and DNA repair in the absence of H2A.X [84]. Altogether, γH2A.X acts as a robust marker of DNA damage and H2A.X is an important facilitator of the chromatin response to DNA damage through its ability to promote the assembly of DNA damage signaling and repair proteins at DNA breaks.

2.4. H2A.X in disease

Although H2A.X is best known for its role in DNA damage signaling and repair, studies have begun to establish the function of this variant in other biological processes. As these have been reviewed extensively, we point readers to these more comprehensive reviews and highlight here a few key studies as examples of the multifunctional nature of H2A.X in mammals [85– 87]. Early work using mouse models of H2A.X loss revealed a tumor suppressor function of H2A.X. Indeed, mice with either one or both copies of H2A.X removed in a p53 mutant background developed tumors and unstable genomes including translocations, establishing H2A.X as haploinsufficient for tumor suppression [46, 88]. Intriguingly, human H2A.X maps to a region frequently altered in several cancers including leukemias, raising the possibility that H2A.X is involved in human cancers [46, 85]. Loss of H2AX results in increased cancer cell migration and invasion, consistent with the notion that H2A.X is involved in the epithelial-mesenchymal transition (EMT) [89]. While loss of H2A.X did not result in increased metastasis in an in vivo model in this study, re-expression of H2A.X resulted in increased tumorigenesis, highlighting a complex relationship between H2A.X and cancer. γH2A.X has also been used extensively to identify DNA damage sites in many cancer cell and tissue samples under normal conditions and following treatments with therapeutic drugs including DNA damaging agents ([90]; reviewed in [91]). This provides a robust readout of DNA damage for analysis of genotoxicity as well as cytotoxicity as apoptotic cells also induce γH2A.X Furthermore, γH2A.X is being evaluated as a potential sensitive and early detection biomarker for cancer and other damage related pathologies including aging, given the high frequency and involvement of genome instability in these diseases.

In addition to cancer, H2A.X has been shown to be involved in other pathologies. H2A.X deficiency in mice results in defective mitochondrial morphology and function, implicating H2A.X in metabolism [92]. Interestingly, H2A.X KO mice display defects in neurological function and behavior that is linked to altered oxidative stress responses [93]. Similar defects have been associated with ATM deficiency in mice and patients with Ataxia-Telangectasia, pointing to a central role for DDR defects in common neurodegenerative disorders. We refer the reader to a recent comprehensive review on this topic . Combined loss of ATM and H2A.X results in embryonic lethality in mice, revealing a genetic dependency for both of these genes during embryogenesis [94]. H2A.X protein levels are reduced in cells exposed to persistent oxidative stress, which alters DNA integrity pathways under such conditions [95]. Together with its role as a modulator of BER [66], this observation has important implications for H2A.X regulation and functions in human pathologies where oxidative stress is implicated, including cancer, aging, neurodegeneration, metabolic disorders and others (reviewed in [96]). Functional roles in meiosis, development, stem cell biology and cell cycle control have also been reported (reviewed in [87]). In many cases, it is unclear how these H2A.X functions are orchestrated, and if they are related to its role in DNA damage signaling and repair. While potential reader proteins of γH2A.X and other PTMs on this variant may participate, little is known about bona fide effector proteins, genome localization, chromatin structure and function involving H2A.X in these biological processes. In many cases, a role for H2A.X in gene regulation rather than in signaling has not been ruled out. Although many questions remain, these studies emphasize the multifunctional and pivotal roles of H2A.X in not only in the DDR but also in other biological processes.

3. H2A.Z

3.1. H2A.Z and genome integrity

The histone H2A variant H2A.Z is encoded by two genes whose protein products, H2A.Z.1 and H2A.Z.2, differ by only 3 amino acids (Fig. 1b). H2A.Z.2 is found in two splice variants with H2A.Z.2.2 containing a truncated and variable C-terminus [97]. Unlike other H2A variants, H2A.Z is essential in mice, where it regulates development and stem cell functions [98]. H2A.Z is known to play multiple roles in transcription, including gene expression, RNA Polymerase II (Pol II) dynamics, enhancer function and 3D genome organization (reviewed in [99]). Not surprisingly, this variant has been implicated in other DNA-related processes including DNA repair, replication, cell cycle and chromosome segregation, and cellular senescence . Many of these pathways are linked to human pathologies and accordingly, H2A.Z has been implicated in cancer, metabolism and neurodegeneration [2, 99]. Given the diverse functions of H2A.Z, it is not unexpected that this variant is also highly modified by PTMs that participate in these underlying biological processes (Fig. 2) [2]. Structurally, H2A.Z incorporation results in nucleosome destabilization when compared to H2A and other variants, suggesting that this H2A.Z property may be important for its unique functions [97, 100, 101]. A detailed overview of H2A.Z biology can be found in a recent review by Colino-Sanguino et al. [99]. We will therefore limit this section to a brief synopsis of key operations performed by H2A.Z in genome maintenance.

H2A.Z is an established contributor to genome integrity in mammalian cells. Early studies revealed that H2A.Z was incorporated at DSBs by the p400 remodeler where it promoted chromatin opening that supported the localization of DNA repair factors Ku70/80 and BRCA1 [102]. Moreover, the FRRUC ubiquitin ligase complex facilitates H2A.Z loading at DNA breaks in a H2A119ub and PARP-dependent manner [103]. In addition to the repair process per se, H2A.Z loading also promotes break-proximal transcriptional repression, a function governed by several chromatin modifying and binding proteins including PRC1, PRC2 and others [104–106]. Interestingly, H2A.Z recruitment to breaks was found to be transient, involving its active removal by the chaperone Anp32E and chromatin remodeler INO80, which in turn was demonstrated to support DSB repair [107, 108]. Remarkably, in the context of HR, depletion of H2A.Z restored the repair defects in Anp32E and INO80 deficient cells, pointing to H2A.Z removal as the main function of these factors in HR. Whether this function is exclusively break-proximal or involves gene regulation or chromatin organization at sites other than DNA lesions is not known.

Although these studies have established mammalian H2A.Z as a contributor to DSB repair, several additional questions remain. It is unclear if H2A.Z plays a structural role in regulating chromatin dynamics at DNA breaks that facilitate repair, or if effector proteins that bind to H2A.Z also participate in repair. An attractive model is that the removal of H2A.Z from breaks may allow other H2A variants to be incorporated at sites of damage, for example macroH2A variants which show delayed accumulation at DSBs [109]. The involvement of mammalian H2A.Z in genome integrity pathways in addition to DSB repair are poorly understood. In yeast, loss of H2A.Z (Htz1) in replication checkpoint mutants results in genome instability and replication stress [110]. In mammalian cells, H2A.Z is enriched at replication origins and is required for a subset of early origins to be activated [111, 112]. Interestingly, H2A.Z was shown to promote H4K20me2 through its association with the methyltransferase SUV420H1 [112]. This histone mark is one of the signals read by 53BP1 for its recruitment to chromatin and function at DSBs [113]. Thus, although H2A.Z is undoubtedly involved in genome integrity, more information is needed to mechanistically define these roles and how they may relate to human deficiencies that potentially involve H2A.Z impairment.

4. macroH2A

4.1. macroH2A genes and functions

Macro-histones are unique among H2A variants in that they possess a large extranucleosomal domain consisting of a disordered linker region and a macrodomain (Fig. 1b). Macrodomains fold into a globular mixed α-helix and β-sheet structure containing a deep groove, which serves as a potential ligand-binding pocket for various forms of ADP-ribose (ADPR), including mono-ADPR, poly-ADPR (PAR), poly(A), and O-acetyl-ADP-ribose (OAADPR). Substantial sequence variation between different macrodomains may explain their ligand-selectivity. While macrodomains are highly evolutionarily conserved, macrodomain-containing histones are predominantly found in vertebrates and notably absent in common model organisms such as yeast and flies. Macro-histones exclusively involve the H2A core histone, and much has been learned in recent years about mammalian macroH2A.1 (reviewed in [2, 114]). Of emerging importance are the distinct biological roles of its two alternatively spliced macroH2A.1 isoforms, macroH2A1.1 and macroH2A.1.2 (Fig. 1b; [115]). The two isoforms differ in 33 aa within the macrodomain, which results in a macroH2A.1.1-specific ability to bind ADPR derivatives, while no ligands have been identified to date for the macrodomain of macroH2A.1.2. Significantly less is known about the independently encoded macroH2A.2 gene, which, like macroH2A.1.2, is unable to bind ADPR [116].

Macro-histones were originally identified as epigenetic modulators of gene expression, regulating developmental and/or senescence-associated transcription programs (see [2, 114] for a comprehensive discussion of this aspect of macroH2A1 function). Reminiscent of its alternatively spliced isoforms, macroH2A1 was shown to localize to two functionally distinct chromatin subtypes marked either by the repressive H3K27me3 mark or by H2B acetylations [117, 118]. However, a formal link between macroH2A.1 splicing and chromatin state remains to be established. Genome-wide macroH2A.1 mapping studies consistently show low to modest enrichment across large chromatin domains, often coinciding with H3K27me3, and to a lesser extent H3K9me3. Association with repressive chromatin marks was shown to facilitate the interaction of heterochromatic regions with lamin-associated domains (LADs), pointing to a role for macroH2A.1 in nuclear organization. Consistent with this, macroH2A.1 loss resulted in perinuclear heterochromatin loss and nuclear membrane deformation [119], which are generally associated with increased DNA damage and genome instability [120, 121]. Extending their contribution to genome maintenance, recent work points to a central and perhaps functionally related role for macroH2A.1.1 and macroH2A.1.2 in DNA repair, while defects in either variant have been linked to increased DNA damage accumulation, altered cell growth, and genomic aberrations [115]. Best understood to date is the role of macroH2A.1 proteins in DSB repair, although increasing evidence points to additional roles in the resolution of other types of DNA lesions. While transcriptional control of repair-relevant genes by macroH2A.1 cannot be discounted, it is the direct recruitment to or presence of macroH2A.1 at DNA lesions that appears to be critical for its DNA repair function(s). In the following, we review recent advances in our understanding of the often splice variant-specific roles of macroH2A.1 in mammalian DNA repair, their impact on genomic and overall nuclear integrity, and discuss possible implications for human pathologies where appropriate.

4.2. DSB repair and macroH2A

The first evidence for macroH2A.1 accumulation at a DNA lesion dates back over a decade, when the macroH2A.1.1 variant was found to accumulate at sites of laser-induced DNA damage (Fig. 4a) [122]. Recruitment of macroH2A.1.1 was dependent on both PARP1 and the ADPR-binding residues in the macroH2A.1.1 macrodomain, suggesting that this process occurs via and/or in response to PAR accumulation at sites of DNA damage. A similar phenomenon was observed at enzymatically induced DSBs [123]. In both cases, macroH2A.1.1 was found to affect DNA damage signaling and repair. While macroH2A.1.1 overexpression resulted in impaired recruitment of the NHEJ effectors Ku70/Ku80 and a concomitant delay in DNA damage resolution [122], macroH2A.1.1 knockdown caused a reduction in 53BP1 recruitment and a moderate increase in NHEJ efficiency [123]. Thus, increased and decreased macroH2A.1.1 levels appear to have a similar impact on the NHEJ process, although the mechanistic basis for this observation remains to be determined. More recently, macroH2A.1.1 was shown to promote repair via Alt-EJ in a splice variant-specific knockout mouse model [124]. Both Ku70/Ku80 and 53BP1 oppose Alt-EJ, consistent with the notion that macroH2A.1.1 may contribute to this repair pathway at least in part by affecting the recruitment of NHEJ effectors. In addition, macroH2A.1.1 was shown to interact with two key Alt-EJ mediators, PARP1 [125, 126] and Ligase 3, in a PARP1 activity-dependent manner that likely involves ADPR binding [124].

Fig. 4.

Involvement of macroH2A.1 in genome maintenance. (a-d) macroH2A.1 accumulation on chromatin coincides with DNA damage-susceptibility. (a) DSBs promote macroH2A.1 accumulation to modulate DSB repair factor engagement and transcription repression at breaks, which in the case of the macroH2A.1.1 variant can involve PARylation. (b) CFSs contain difficult to replicate elements including G4 structures, heterochromatin and repetitive DNA, which are often enriched for macroH2A.1 that in turn protects from replication stress and DNA damage. (c) rRNA-encoding rDNA repeat segments are enriched for macroH2A.1, which suppresses rDNA transcription and may protect from transcription:replication conflicts and resulting RNA:DNA hybrids. (d) MacroH2A.1 enrichment at subtelomeric DNA counteracts replication stress and DNA damage formation. Whether or not macroH2A.1 suppresses TERRA transcripts and subsequent RNA:DNA hybrids remains to be determined. macroH2A.1-containing nucleosomes are in light red, red lines depict macroH2A.1 enrichment and distribution based on ChIP.

Like macroH2A.1.1, macroH2A1.2 is recruited to sites of DSBs, albeit with delayed kinetics and in a manner dependent on ATM signaling rather than PARP1 activation [109]. MacroH2A.1.2 was found to be required for optimal BRCA1 recruitment and consequently promoted HR, suggesting that the two splice variants may support distinct DSB repair outcomes. While depletion of either variant had a similar, modest positive effect on NHEJ [109, 123], opposite effects were indeed observed for both HR and Alt-EJ upon variant-specific macroH2A.1 depletion [109, 124]. These findings support the view that macroH2A.1 variants regulate DNA repair pathway choice specifically at resected DNA ends, the obligate intermediates for both HR and Alt-EJ. Given that macroH2A.1.2 facilitates end resection via the BRCA1/CtIP axis [109], it is tempting to speculate that its absence results in limited resection, which impairs HR but not Alt-EJ [127]. MacroH2A.1.1, in turn, facilitates recruitment of the Alt-EJ machinery when macroH2A.1.2 abundance is low and resection and/or HR factor recruitment is suboptimal. While it is plausible that competition amongst macroH2A.1 variants during repair of resected ends occurs in a genome-site specific manner, a more general effect at the level of overall protein abundance cannot be excluded. The implications of these findings for genome integrity and ultimately malignant transformation are discussed in the context of fragile DNA elements below.

The presence of macroH2A.1 at DSBs does not only affect repair factor recruitment, but further has a significant impact on the DSB-surrounding chromatin environment, although the two processes are likely closely intertwined. Both macroH2A.1.1 and macroH2A.1.2 have been shown to promote a dynamic condensation of break proximal chromatin [109, 122], consistent with their established role in repressive chromatin formation [2]. If and how this structural change in break-proximal chromatin contributes to repair outcomes remains to be established. Of note, chromatin decondensation through histone deacetylase inhibition resulted in impaired BRCA1 retention, similar to the effect of decondensation observed upon macroH2A.1.2 depletion [109]. However, increased acetylation of specific histone residues that interfere with 53BP1 chromatin association has been linked to an increase in BRCA1 recruitment [128]. Together, these findings point to a complex role for chromatin dynamics in repair factor assembly and disassembly that is likely directly affected by the DSB-surrounding chromatin environment.

Consistent with a preference for open chromatin, HR has been shown to be preferentially active at transcribed loci, likely to avoid mutagenic repair of transcribing genic regions [129]. A set of recent observations involving macroH2A.1 among other factors, may help reconcile the seemingly conflicting requirements for active and repressive chromatin in the context of HR. The bromodomain-containing and hence acetyl-histone-binding ZMYND8 protein is recruited to breaks in actively transcribed genes, which in turn initiates the recruitment of the NuRD (nucleosome remodeling and histone deacetylation) complex and establishment of a repressive chromatin environment at genic DSBs [130–133]. ZMYND8/NuRD accumulation is further dependent on PAR-mediated recruitment of the histone demethylase KDM5A and concomitant H3K4 demethylation [131, 134]. Of note, optimal KDM5A recruitment also required the presence of macroH2A.1.2 [134], suggesting a need for the dynamic exchange of active and repressive chromatin marks to facilitate an HR-permissive repair environment. Consistent with this, DBSs were shown to induce sequential expansion and recondensation of chromatin, the latter of which being dependent on macroH2A.1.2 [109, 135]. Although not formally demonstrated, H2A.Z may account for the former, which could at least in part explain its role on HR (see above). Dynamic chromatin condensation further provides a plausible mechanistic basis for previously reported DSB-induced transcriptional silencing thought to help restrict gene expression until repair is completed [104–106]. Indeed, depletion of KDM5A, ZMYND8 or macroH2A.1.2 each results in defective break-proximal transcriptional silencing as well as impaired HR [20, 130, 131, 134]. If repression of DSB-proximal transcription is a prerequisite for HR or an auxiliary event to insulate the repair environment from active RNA polymerase encounters remains to be established.

It should be noted that whether or not macroH2A.1 variants are incorporated into break-proximal nucleosomes is an ongoing debate. While PARP1-dependent macroH2A.1.1 accumulation at DSBs does not appear to involve its nucleosome deposition [122, 123], recent work in the context of replication stress-associated DNA damage points to chromatin incorporation of macroH2A.1.2 [82]. However, no definitive histone chaperones have been identified for either variant. Moreover, given that both PAR and intrinsically disordered domains can promote liquid:liquid phase separation, it is possible that macroH2A.1 variant accumulation may reflect an aspect of the recently reported liquid condensate formation at DSB sites through either their disordered linker region or macroH2A.1.1:PAR binding (Fig. 4a; [11, 136, 137]). While a role for macroH2A.1.1 in HR remains to be determined, early PAR/macroH2A.1.1 accumulation may reflect an effort to restrain HR until cells encounter a more HR-permissive environment. A similar role has recently been proposed for phase separation of DNA lesions encountered in mitosis [11, 138]. Irrespective of the precise nature of macroH2A.1 variant accumulation at DSBs, the body of work discussed here underlines the importance of macroH2A not only in DSB repair but also for the coordination of the dynamic structural changes to chromatin that are associated with this process.

4.3. Involvement of macroH2A In ssDNA lesion repair?

In contrast to DSB repair, little is known about how macroH2A.1 affects the repair of ssDNA breaks, although the prominent role for PARP1 in the cellular response to ssDNA breaks points to involvement of at least the PAR-binding macroH2A.1.1 variant. In support of the latter, macroH2A.1.1 was recently shown to protect cells from oxidative DNA damage in a manner dependent in part on the activity of PARP1 [126]. Depletion of macroH2A.1.1 resulted in a pronounced increase in necrotic cell death, which was attributed to defective PAR chain stabilization by macroH2A.1.1, a feature that has been linked to excessive nuclear NAD⁺ consumption and metabolic homeostasis [126, 139]. Moreover, macroH2A.1 was able to facilitate the repair of the predominant oxidative DNA lesion 8-oxo-guanine, although the underlying mechanism remains unclear [126]. Support for a role in BER, the predominant repair pathway for 8-oxo-guanine, comes from the recent observation that macroH2A.1.1 can bind Ligase 3, which in addition to Alt-EJ is a key effector of BER [124]. Of note, macroH2A.1.2 depletion also results in an increase in susceptibility to oxidative stress, albeit in a PARP1-independent manner [126]. It will be interesting to determine if the latter is due to DSBs resulting from excessive oxidative damage, or if it reflects a bona fide role for macroH2A.1.2 in BER, perhaps complementary to or synergistic with that of macroH2A.1.1. Lastly, it will be imperative to assess if macro-histones, as well as other variants, are involved in additional ssDNA repair processes.

4.4. macroH2A and replication stress

Both ssDNA and dsDNA breaks are a hallmark of DNA replication stress. Of note, macroH2A.1 mapping efforts across a range of mammalian cell lines and tissues has revealed preferential macroH2A.1 enrichment at expansive, replication stress-susceptible genomic regions, including common fragile sites (CFSs), the inactive X chromosome, telomeres, and nucleolar ribosomal DNA (rDNA); (Fig. 4b–d; reviewed in [114]). Moreover, macroH2A.1 protects at least a subset of these regions from DNA damage, which has been most comprehensively studied at CFSs [82]. CFSs are genomic regions rich in secondary DNA structures such as G-quadruplexes, stem loop-prone short and long interspersed nuclear elements (SINEs/LINEs), RNA:DNA hybridforming regions and densely packed heterochromatin, all of which pose obstacles to DNA polymerases [140]. Stalling of the replication machinery results in the accumulation of under-replicated ssDNA regions, which activates a DNA damage checkpoint response involving ATR and CHK1 kinases to delay replication until the resulting damage can be resolved or repaired [141]. Consistent with its role at DSBs [109], macroH2A.1.2 facilitates the recruitment of the replication fork-protective BRCA1 protein to stalled replication forks [82]. MacroH2A.1.2 depletion conversely results in the accumulation of various hallmarks of replication stress, including phosphorylation of the ssDNA sensor and ATR activator RPA, accumulation of FANCD2, a marker of under-replicated CFSs, impaired RAD51 loading and concomitant replication fork degradation, suggesting HR as the main effector of macroH2A.1-mediated fork protection [82, 142]. Of note, CFS-associated DNA damage appears to be a driver of macroH2A.1 deposition, which was proposed to help shape a protective CFS chromatin environment for subsequent cell divisions (Fig. 4b) [82]. Both the histone chaperone FACT and the LSH2/HELLS chromatin remodeler have been implicated in this process [82, 142]. Since LSH2/HELLS is frequently mutated in Immunodeficiency Centromeric Instability Facial Anomalies (ICF) 4 syndrome [143], and given the epistatic function of macroH2A.1 and LSH2 in response to replication stress, it will be interesting to investigate if macroH2A.1 is similarly implicated in this severe human disease with symptoms including immunodeficiency, neurologic defects, and reduced growth [142, 144].

It is worth noting that ChIP-Seq analyses with splice variant-specific antibodies revealed similar enrichment patterns for both macroH2A.1.1 and macroH2A.1.2 [82, 145]. While replication stress has been proposed as a driver for CFS-specific accumulation of macroH2A.1.2, the underlying principles that govern macroH2A.1.1 deposition at these sites remain to be established. Their genomic overlap suggests that much like DNA methylation, existing macroH2A.1 molecules may serve as a bookmark for subsequent macroH2A.1 deposition, irrespective of the isoform. Consistent with this, LSH2 was recently shown to facilitate incorporation of two distinct macroH2A variants, macroH2A.1 and macroH2A.2 [146]. While macroH2A.1.1 does not appear to be directly implicated in the replication stress response [82], it may mediate a complementary genome-protective effect through its ability to inhibit basal poly-ADP ribose polymerase 1 (PARP-1) activity and concomitant nuclear NAD⁺ consumption [139].

The balanced presence and/or activity of both macroH2A.1 variants appears to be essential for genome stability at the replication stress-susceptible inactive X chromosome (Xi), which is significantly enriched for all macroH2A variants [114]. Consistent with its role at CFSs, macroH2A.1.2 loss resulted in a pronounced increase in both aneuploidy and micronuclei formation, which can be attributed to an accumulation of under-replicated DNA causing DNA fragility and DSBs upon entering mitosis [147]. However, while macroH2A.1.2 loss was the likely cause for the underlying lesions, it was their aberrant, Alt-EJ-dependent repair supported through macroH2A.1.1 that accounted for anaphase bridge formation and associated chromosomal aberrations [124]. Supporting the physiological relevance of this finding, mice deficient in macroH2A.1.2 have increased female-biased lethality observed as early as embryonic day 9.5, while mice that lack both macroH2A.1.2 and the Alt-EJ-promoting macroH2A.1.1 variant show no overt survival defects [124, 148].

Although chromosomal defects upon macroH2A.1.2 loss were most pronounced at the macroH2A.1-rich Xi, lower levels of aneuploidy could be observed on autosomes, suggesting that CFSs across the genome may be prone to similar defective repair when macroH2A.1 splicing is perturbed. Aberrant macroH2A.1 splicing is a frequent feature of most cancers and has been linked to poor patient survival [2, 115]. It is, thus, worth considering that macroH2A.1 splice variant profiles may reflect DSB repair potential and therapy responsiveness, given that both HR and Alt-EJ have been identified as cancer-specific vulnerabilities [22, 149–151]. In support of this notion, macroH2A.1 appears to alter tumor progression in a splice-variant and cancer-type dependent manner. Importantly, the impact of macroH2A.1 on cancer progression is likely to reflect a combination of its gene regulatory and genome maintenance functions (reviewed in [2, 114]).

4.5. Protection from replication:transcription conflicts by macroH2A?

CFSs are not the only genomic regions where genome integrity is impacted by macroH2A.1. Both telomeric and ribosomal DNA (rDNA) are enriched for macroH2A.1 and display functional defects upon macroH2A1 loss (Fig. 4c–d; [114]). Notably, at both loci, transcription poses a major source of genome instability, pointing to a common, genome-protective role of macroH2A.1-associated repressive chromatin [152]. In the case of the rDNA, rRNA-encoding genes exist in several hundred copies spread in clusters over 5 chromosomes, which aggregate in the nucleolus. Only a subset of rDNA repeats is expressed at any given time, and the silencing of the remaining rDNA units is at least in part mediated by macroH2A.1 [119, 153]. To prevent transcription:replication conflicts in active rDNA units, evolutionarily conserved replication fork barriers (RFB) separate the rRNA encoding region from the rDNA intergenic spacer region, where DNA replication initiates [154]. Spontaneous DNA breaks have recently been detected precisely at this barrier, even in the absence of added replication stress [155]. Notably, the separation of the rRNA-coding region and the intergenic spacer is recapitulated in the pattern of macroH2A.1 accumulation, which is restricted to the former (Fig. 4c; [153]). While it remains to be determined whether and how macroH2A.1 mediates protection from replication:transcription conflicts, several recent findings support this notion: first, macroH2A.1 was shown to coordinate replication and transcription processes during human papilloma virus replication in a process that appears to involve their spatial separation within viral replication foci [156]; second, macroH2A.1 has been implicated in the suppression of replication stress at telomeres [157], which is at least in part attributed to the transcription of telomeric repeat containing RNA (TERRA) (Fig. 4d; [158, 159]), third, TERRA-associated transcription:replication conflicts and the resulting formation of RNA:DNA hybrid structures can be resolved by BRCA1 [159], raising the possibility that macroH2A.1.2-mediated BRCA1 recruitment may help protect from these often detrimental, aberrant structures. Finally, telomeric RNA:DNA hybrids are particularly abundant in tumor cells undergoing Alternative Lengthening of Telomeres, which were shown to depend on macroH2A.1.2 for efficient tumor growth [157]. It is worth noting that an involvement of macroH2A.1 and/or its alternative splicing in RNA:DNA hybrid resolution more generally could have a significant impact on a range of RNA:DNA hybrid-associated pathologies [160–162], although no evidence for such a function exists to date.

4.6. macroH2A.1 PTMs in repair

Like all H2A variants, macroH2A.1 is subject to PTMs, which may affect its function in the repair processes described above (see Fig. 2 for a summary of known/validated macroH2A.1 PTMs). In support of the latter, macroH2A.1-ubiquitination has been linked to the DDR-associated E3 ligases RNF168 and BRCA1/BARD1 [163, 164]. RNF168 mediates K11Ub of macroH2A.1 in proximity to an alpha1 extension helix region that is conserved among H2A variants [163], and an analogous N-terminal Ub modification is observed at K15 on H2A.Z [163, 165]. Whether or not these modifications are induced in response to DNA damage remains to be determined, as does their impact on DNA repair. However, ubiquitination of the equivalent K residues on H2A plays an important role in the control of 53BP1 and BRCA1/BARD1 repair factor recruitment, directing the choice between HR and NHEJ [20, 52–55, 166]. In further support of a link to HR, BRCA1/BARD1 mediates macroH2A.1 ubiquitination at K123, although the functional relevance of this modification remains to be fully understood [164]. Of potential relevance for macroH2A.1-mediated Alt-EJ and Xi instability [124], phosphorylation of S137 was found to be enriched in mitosis and excluded from the inactive X chromosome [167]. Other macroH2A.1 phosphorylations may further be relevant for splice variant-specific macroH2A.1 functions, as T129 and S170 in the common macroH2A.1 linker region were found to be preferentially modified in macroH2A.1.1 and macroH2A.1.2, respectively [168]. What regulates this differential phosphorylation, and whether or not it contributes to the distinct repair outcomes associated with each variant, remains to be determined.

Altogether, macroH2A.1 plays a central role in maintaining nuclear integrity both in response to defined DNA lesions and at genomic regions that are prone to DNA damage. Remarkably, these protective functions involve several, possibly complementary functional and/or structural aspects of macroH2A.1 biology, including its genomic location, PTMs and the distinct biological roles of macroH2A.1 splice variants.

5. Mechanisms of genome integrity by other H2A variants

5.1. H2A.J

The H2A histone variant H2A.J is encoded by the H2AFJ gene in humans and is highly similar to core H2A, with a difference of approximately 5 amino acids only (Fig. 1b). This includes the presence of an SQ motif in the C-terminus that, like H2A.X, is phosphorylated after DNA damage [169]. H2A.J was first linked to DNA damage when its expression was observed to increase 10-fold in both replicative and DNA-damage induced senescent human fibroblasts [170]. H2A.J was shown to be required for the induction of immune and inflammatory genes, including cytokines associated with senescent-associated secretory phenotype (SASP). The ability of H2A.J to promote inflammatory genes may be contributed to its weakened interaction with linker histone H1. This is in line with the differences between H2A and H2A.J in the C-terminus where H2A binds H1. It has been proposed that phosphorylation of H2A.J could further weaken these contacts with H1 and DNA resulting in increased gene expression of immune and inflammation genes following DNA damage [169]. Of note, macroH2A.1 variants have also been implicated in senescence and the associated SASP phenotype [171]. If and how H2A.J relates to or affects macroH2A.1 function may be a worthwhile subject for future investigation. Further supporting a role in senescence and aging, aged tissues in mouse and human, as well as prolonged irradiation of mouse tissues, resulted in increased expression of H2A.J. In addition, gene amplification and overexpression of H2A.J has been reported in a subset of cancers, consistent with its involvement in DNA damage-associated human pathologies [170]. However, while irradiation increases the expression of H2A.J and leads to its phosphorylation, how these changes contribute to its role in genome maintenance remains unclear. H2A.J phosphorylation was observed only after high levels of radiation in mouse cells and H2A.J localization at breaks was not observed after acute irradiation in human cells [169, 172]. Thus, several questions remain for the potential involvement of H2A.J in genome integrity pathways. The identity of the kinase(s) involved in DNA damage-dependent phosphorylation of this variant on S123 are unknown, as is the localization and presence of this modification in human cells. Potential readers of this damage-induced mark are as-yet-unidentified and roles for H2A.J beyond those reported for gene regulation in response to damage remain undetermined. It will be informative to perform additional experiments to address these questions as they may reveal the function of this histone variant in damage-associated human diseases including cancer, aging and potentially neurodegeneration, a condition where H2A.J has not yet been studied. A recently reported H2A.J knockout mouse is viable and fertile and may provide an important next step to dissecting the physiological impact of H2A.J deficiency, particularly under conditions of stress [173].

5.2. H2A.B

H2A.B (also referred to as H2A.BbD) is another histone H2A variant implicated in DNA damage responses. This H2A variant is only 48% identical to H2A and was first identified as being excluded from the inactive X-chromosome in female human cells, suggesting a role in gene regulation [174]. Later work showed that H2A.B nucleosomes display reduced spacing compared to H2A and localize to active genes [175]. Depletion of H2A.B resulted in alterations in gene expression and defects in splicing. H2A.B is further likely to be actively involved in the response to DNA damage, as overexpressed, GFP-tagged H2A.B localizes to sites of laser microirradiation and replication. Reduced S-phase progression and defective UV damage repair was observed upon overexpression of H2A.B [176]. Proteomic analysis of this variant revealed an association with replication proteins, which is in line with H2A.B being involved in DNA replication and repair. Thus, several histone H2A variants in addition to H2A.X, H2A.Z and macroH2A also participate in genome integrity pathways. These variants remain insufficiently studied and several important questions remain. Neither do we know what cellular context these variants function in, nor are the molecular machinery regulating the dynamics, potential modifications and interaction partners identified or any disease relevance defined in most cases. Addressing these questions will likely provide important insights into the collective role that histone H2A variants play in genome integrity.

6. Conclusions and Future Directions

H2A variants come in many varieties and account for the majority of the nucleosome diversity observed in mammalian cells (Fig. 1a). Remarkably, all H2A variants studied in-depth to date show some involvement in genome maintenance pathways, underlining the importance of the chromatin environment for genomic integrity. While several H2A variants, like H2A.X, H2A.Z and more recently macroH2A.1, have been studied extensively, the relevance of variants such as H2A.J and H2A.B is only emerging. Moreover, changes between annotated H2A variants and core H2A are sometimes limited to a few amino acids, analogous to the variability observed between core H2A encoding genes, which draws into question what defines a histone variant. It will be interesting to determine if coding sequence variation across the many core H2A genes translates into functional differences, as observed for example between H2A.J and H2A.

Building on what we have learned to date, several key questions are emerging that will require attention if we want to truly understand how diversity in H2A histones, and nucleosome composition more generally, can be leveraged to understand repair pathways, genome integrity, and ultimately human diseases and their respective therapies. For one, the often pronounced, differential impact of H2A variants on DNA repair outcome and genome stability raises the question of what controls their incorporation and dynamic exchange in the context of distinct DNA lesions and chromatin states. Chaperones of key modulators of DNA repair such as macroH2A.1 remain to be uncovered, and the contribution of H2A variant exchange to repair kinetics and progression is only beginning to be understood. Is nucleosome incorporation even critical in all cases or are many of these functions channeled through pre-existing variant deposition as well as the numerous PTMs identified for H2A variants? Moreover, at least for macroH2A.1, chromatin association rather than nucleosome incorporation may be sufficient for their function in repair. What is the impact of non-break-proximal H2A variants on DNA repair? Recent insights into the importance of three-dimensional chromatin configuration for DSB repair suggest that TAD formation, chromatin looping and possibly phase separation can define repair outcome [11]. These processes are likely to be modulated by H2A variant nucleosomes, even when they are not directly recruited to or present at DNA lesions. It will further be important to functionally dissect the relevance for TADs and other three-dimensional structures for the DNA repair process itself. Manipulation of TAD-associated histone variants, such as H2A.X, may present a promising approach to address this issue. Moreover, there are many H2A variant PTMs with yet to be defined contribution to DNA repair. Given the significant impact that small changes in H2A variant modifications can have on repair outcomes and genome integrity, a wealth of additional information can be expected from further studies of PTM dynamics and the underlying modifiers and readers.

Finally, disease relevance of H2A variants remains an underexplored area of investigation. Linking H2A variants and their modifications and/or modifiers to disease-associated mutations should be a focus of future research. This will greatly benefit from the wealth of information that is being generated through various (single-cell) omics efforts to characterize human tissues, such as the Human Cell Atlas and the cancer-focused Human Tumor Atlas Network and Clinical Proteomic Tumor Analysis consortium [177–179]. Emphasis should also be put on regulators of alternative splicing, which may be of particular relevance for macroH2A.1 and H2A.Z physiology and function. Ultimately, a greater understanding of how H2A variants participate in the various DNA-templated processes including transcription, replication and DNA repair will be required to define the contribution that these epigenetic regulators have in normal physiology and disease. This information has the potential to be leveraged for new therapeutic strategies and tools such as biomarkers for human diseases, including those impacted by the loss of genome integrity such as aging, neurodegeneration and cancer.

Highlights.

• Genome integrity relies on diverse histone H2A variants and their modifications

• γH2A.X foci represent 3-D damage signaling frameworks formed by chromatin loops

• macroH2A.1 is a splicing-regulated modulator of DSB repair pathway choice

• H2A.J and H2A.B variants are emerging as effectors of DNA repair

• H2A variant functions including genome integrity are linked with several human pathologies