The tale of a tail: histone H4 acetylation and the repair of DNA breaks

Abstract

The ability of cells to detect and repair DNA double-strand breaks (DSBs) within the complex architecture of the genome requires co-ordination between the DNA repair machinery and chromatin remodelling complexes. This co-ordination is essential to process damaged chromatin and create open chromatin structures which are required for repair. Initially, there is a PARP-dependent recruitment of repressors, including HP1 and several H3K9 methyltransferases, and exchange of histone H2A.Z by the NuA4-Tip60 complex. This creates repressive chromatin at the DSB in which the tail of histone H4 is bound to the acidic patch on the nucleosome surface. These repressor complexes are then removed, allowing rapid acetylation of the H4 tail by Tip60. H4 acetylation blocks interaction between the H4 tail and the acidic patch on adjacent nucleosomes, decreasing inter-nucleosomal interactions and creating open chromatin. Further, the H4 tail is now free to recruit proteins such as 53BP1 to DSBs, a process modulated by H4 acetylation, and provides binding sites for bromodomain proteins, including ZMYND8 and BRD4, which are important for DSB repair. Here, we will discuss how the H4 tail functions as a dynamic hub that can be programmed through acetylation to alter chromatin packing and recruit repair proteins to the break site.

This article is part of the themed issue ‘Chromatin modifiers and remodellers in DNA repair and signalling’.

1. Introduction

Maintaining the integrity of cellular DNA is a critical process in mammalian cells. Cells are constantly exposed to genotoxic events which cause oxidative damage to sugar and base residues, the formation of DNA adducts or even the generation of DNA breaks. As a result, cells have evolved multiple DNA repair pathways that constantly monitor the genome and that can act to rapidly repair these potentially mutagenic DNA alterations. A key element of DNA repair is that detection of damaged DNA leads to the recruitment of specific DNA repair complexes to the damaged chromatin, thereby facilitating repair. Consequently, DNA repair pathways are highly integrated with, and dependent on, the cellular pathways that modulate chromatin organization [1–3]. In particular, the repair of DNA double-strand breaks (DSBs), in which the DNA helix is severed, represents a particular challenge for the DNA repair machinery. Work from many laboratories has now demonstrated that repair of DSBs depends on the pre-existing chromatin organization, the recruitment of specific chromatin remodelling complexes to the DSB and dynamic changes in nucleosome positioning, modification and histone composition, and these subjects are addressed in several reviews in this issue [4–9]. Of particular importance to DSB repair is the need for cells to create an open, relaxed chromatin structure which favours DSB repair. Here, we will examine how highly focused changes in acetylation of histone H4 play a critical role in directing changes both in chromatin organization and in promoting recruitment of DSB repair proteins to sites of DNA damage.

2. Double-strand break repair: the basics

The initial cell-based response to DSBs is well defined and has been covered in detail in other recent reviews [10–13]. DSBs are recognized by the multifunctional MRE11-RAD50-NBS1 (MRN) complex. MRN recruits the ATM kinase to breaks, activating ATM's kinase activity [10] and phosphorylating multiple proteins [14], including the histone variant H2AX [15]. MDC1 then binds to phosphorylated H2AX (γH2AX) [16] which, in turn, creates a platform for activated ATM, facilitating spreading of H2AX phosphorylation for hundreds of kilobases along the chromatin [1,10,11,17]. MDC1 also recruits the ubiquitin-ligases RNF8 and RNF168, which can ubiquitinate both chromatin [11] and histones H2A/H2AX [18,19]. This ubiquitination is essential for recruitment of 53BP1, a dual reader that interacts with ubiquitinated H2AK13/15 (created by RNF168) and H4K20me2 [19]. 53BP1 recruitment plays a pivotal role in directing DSBs to either NHEJ (NHEJ) or homologous recombination (HR) repair pathways.

In non-homologous end-joining, damaged DNA ends are recognized by the Ku70/80 complex and DNA-PKcs, followed by recruitment of nucleases which minimally process the damaged ends and promote religation by DNA ligase IV [20]. NHEJ therefore has low fidelity and can create mutations/deletions/insertions at the site of damage. HR-mediated repair operates during S-G2M and uses the sister chromatid generated during replication as a template for DSB repair. HR-mediated repair requires end-resection of the DNA (by CtIP and other nucleases [21]) to produce a single-stranded (ssDNA) molecule, which then invades the sister chromatid to locate homologous DNA sequences that can prime repair [22]. HR is therefore a high-fidelity repair mechanism, but is effectively limited to operation only during the S-G2M part of the cell cycle.

It is now clear that the ability of cells to repair DNA breaks by either NHEJ or HR is dependent on the chromatin architecture surrounding the area of damage. Gaining access to breaks and processing the damaged area require significant reorganization of the chromatin template to allow the DNA repair machinery to carry out its job. Processes such as end-resection of the DSB and homology searching during HR require significant remodelling of both the damaged chromatid and the adjacent sister serving as the template for repair. The repair of DSBs is therefore intimately linked with dynamic changes in the underlying chromatin and nucleosome architecture at the site of damage, indicating the key role that histones and nucleosomes have in determining the outcome to DSB repair. The contribution of chromatin remodelling complexes to DSB repair has been extensively covered in many excellent recent reviews [3,4,23,24], including several that accompany this article. However, the role of histone modifications during DSB repair, including acetylation of the N-terminal tail of histone H4, has received less attention. H4 acetylation (H4Ac) plays a central role in regulating chromatin compaction, loading of the 53BP1 protein and directing DSB repair. In this review, we will focus on understanding how dynamic changes in the acetylation of the histone H4 tail contribute to DSB repair.

3. Histone modifications and regulation of chromatin

DNA transactions such as replication, transcription and DNA repair are tightly regulated by the local chromatin conformation. The basic unit of chromatin is the nucleosome, which is made of approximately 147 bp of DNA wrapped around a histone octamer. The four histones (H2A, H2B, H3 and H4) form H2A-H2B and H3-H4 dimers, with a complete nucleosome containing two H2A-H2B dimers and two H3-H4 dimers. There are also specialized histone variants, including H2AX [17], which is required for DSB repair in heterochromatin, and H3.3 and H2A.Z [25–27], which are important for transcription and organization of specific chromatin domains. Histones have a common structure, with the central histone-fold region forming the core of the nucleosome, and unstructured N- and C-terminal tails that extend outwards from this. The N-terminal tails of histones contain well-conserved lysine residues which can be modified at the ɛ-amino group. These lysines can be mono-acetylated by a large family of lysine acetyltransferases (KATs) [28,29], mono-, di or tri-methylated by lysine methyltransferases (KMTs) or subject to ubiquitination or sumoylation by ubiquitin and SUMO ligases [11]. Many of these histone modifications, including H3K9 methylation [30–33], increased sumoylation and ubiquitination [11], and increased acetylation of histone H4 [34–40] play key roles in DSB repair.

Increased histone acetylation, and in particular, increased acetylation of the H4 tail (H4Ac) at K16, is strongly linked to open, transcriptionally active regions of the chromatin [41,42]. H4Ac can alter chromatin organization through several mechanisms. (i) Charge neutralization: Lysine acetylation is carried out by a family of KATs that transfer the acetyl group from acetylCoA to the ɛ-amino of lysine, neutralizing the positive charge on lysine. This leads to a net reduction in the positive charge on histones, which can reduce charge-regulated histone–DNA interactions as well as reduce interactions between neighbouring nucleosomes [43–46]. Further, specific acetylation at H4K16 inhibits interaction between the H4 tail and a regulatory domain on the surface of the nucleosome called the acidic patch [43,47]. The acidic patch is formed by a concentration of acidic residues (provided by histone H2A and residues from H2B [48]) that cluster together on the nucleosome's surface. These residues form a shallow, acidic groove to which proteins such as HMGN2 [49], the KSHV LANA protein [50], the PRC1 ubiquitin ligase complex [51] and the unacetylated H4 tail can bind. Binding of the H4 tail on one nucleosome to the acidic patch on an adjacent nucleosome promotes the formation of packed nucleosomal arrays [43,52]. However, acetylation of H4 (at lysine 16) blocks H4 tail binding to the acidic patch [1,43,48,52–54], preventing compact chromatin fibre formation, increasing nucleosome mobility and promoting unstacking of packed nucleosome arrays. Acetylation of the H4 tail can therefore shift chromatin and nucleosome packing to create more open, flexible structures. As we will discuss later, the acetylation of H4 during DSB repair is a key factor in limiting H4 binding to the acidic patch and creating open, flexible nucleosome structures that favour repair [1,34]. (ii) Acetylation limits further modification. Because lysine residues can be subjected to multiple modifications, acetylation can prevent other modifications on the same residue. For example, H3K9Ac, a modification associated with transcription, can prevent H3K9 methylation, and limit DNA repair events dependent on H3K9me2/3 [55]. Acetylation of H2A can prevent damage-dependent increases in H2A ubiquitination at the same lysine [37], limiting both 53BP1 loading and NHEJ and thereby favouring repair by HR. This ability of individual lysine residues to be acetylated, methylated or ubiquitinated provides a powerful mechanism for precise cross-regulation of histone function and recruitment of specific proteins to the nucleosome. (iii) Protein modules that bind acetylated lysine. Acetylation of histones also creates binding sites for a large family of chromatin proteins that contain bromodomains (BRDs) [56]. BRDs are acetyl-lysine readers that function in diverse processes, including transcriptional regulation and DNA repair. Many BRD proteins function as platform proteins that recruit other complexes and activities, such as HDAC activity or transcriptional co-activators, to genes. Further, proteins containing BRD domains often contain other histone modification readers, such as PHD domains, which can allow these proteins to read multiple histone marks. Several BRD proteins are recruited to DSBs, including BRD4, ZMYND8, ACF1, TRIM28 (KAP-1) and TRIM33 [57–61], and KATs such as p300 and GCN5 (reviewed in [28]), although not all of these proteins require the BRD for retention at DSBs [62]. In the next section, we will examine how these three functions of lysine acetylation (charge neutralization, blockade of modification and recruitment of BRD proteins) contribute to the altered chromatin organization after DNA damage and to the repair of DSBs.

4. Double-strand break repair and chromatin reorganization

Exposure of cells to DNA damage is strongly associated with a decrease in chromatin compaction. Many types of DNA damage [63], including ionizing radiation (IR) [64,65], lead to increased sensitivity of DNA to digestion with nucleases, indicating increased access to the linker DNA between nucleosomes. Further, IR decreases histone–DNA interactions, leading to an increase in NaCl solubility of histones [34,35,66]. In heterochromatin, a dedicated mechanism is required to promote chromatin decompaction and DSB repair. This involves rapid phosphorylation of the KAP1 repressor (by ATM [64,65]) and ejection of the repressive CHD3 complex [65,67,68], which function to provide access to and repair of DSBs in heterochromatin [69]. Indeed, recent work indicates that heterochromatic DSBs may be moved to the periphery of the heterochromatin for repair [5,70–73], underscoring the need for large-scale chromatin decompaction and mobility during DSB repair. Use of microscopy has also shown that regions undergoing DNA damage exhibit a rapid (seconds–minutes) expansion of the local chromatin after DNA damage [74–76]. Further, these shifts to a more open, mobile chromatin structure are dependent on increased chromatin acetylation, with increased acetylation of the histone H4 tail being of key importance [34–36,38,39,66,77,78].

Although acetylation of H4 may drive the formation of open chromatin, H4 acetylation is only one of many events that are required to create open, relaxed chromatin structures at DSBs. Paradoxically, one of the earliest events after DNA damage is the rapid accumulation of multiple repressive proteins at the damage site. This accumulation of repressive proteins (figure 1) requires polyADP-ribose polymerase (PARP) family members, which are recruited to DSBs and which rapidly PARylate the chromatin [79]. PARylation recruits a wide range of repressive proteins to the chromatin, including complexes containing HP1 and KAP-1 [80–82], the H3K9 methyltransferases SUV39H1 [30] and PRDM2 [31], and several macroH2A variants [31,83]. The NuRD complex, a multisubunit repressor complex containing the CHD3/CHD4 chromatin remodellers and HDACs, is also loaded onto the chromatin during this time [57,74,75,84,85]. This rapid concentration of repressor complexes such as HP1 onto the damaged chromatin, coupled with local increases in H3K9 methylation [30] and HDAC activity [75,86,87], suggests that there is an initial formation of repressive chromatin structures directly at the DSB, rather than decompaction. However, because PARylation at damaged chromatin is transient, these repressive complexes are only retained at the break for a few minutes before they are lost. This removal is facilitated by rapid removal of PAR chains by PARG, phosphorylation of KAP1 (by ATM [30]) and recruitment of histone demethylases such as KDM4A [88]. This has led to the idea (figure 1) [1,89] that the initial response to DSBs is the rapid, PARP-dependent accumulation of repressive complexes at the site of damage. These complexes may repress local transcription and prevent RNA Pol II movement through damaged chromatin [90–93]. Further, the presence of HDACs and remodelling ATPases [57,75,94] may function to rewrite the local epigenetic code in preparation for repair. It is also possible that these repressive structures may temporarily limit the mobility of the damaged chromatin, maintaining the DNA ends at the break in close proximity prior to initial processing of the damage. Overall, this suggests that the initial response to DNA damage is rapid loading of repressive factors, followed by removal of repressors and transition to a more open, repair-favourable chromatin. However, some studies indicate alternative interpretations. Hinde et al. [95], using high-resolution imaging, demonstrated that the chromatin at the site of DNA damage becomes more accessible, whereas the surrounding chromatin becomes more compact. The rapid accumulation of repressive factors may therefore be occurring on the surrounding chromatin, creating a more compact chromatin region surrounding the locally damaged region, which enjoys increased mobility. Although ChIP studies have noted the presence of HP1 and H3K9me3 within 1 kb of the break [30], it remains possible that many of the repressive factors identified by microscopy studies reflect increased accumulation at sites distant from the DSB, rather than directly at it. Finally, it has also been noted that condensed chromatin structures may occur during DSB repair processing through accumulation of macroH2A1 at DSBs [31], indicating that dynamic shifts between open and compact chromatin may occur as the cell repairs damaged DNA.

Figure 1.

Temporal ordering of nucleosome dynamics during DSB repair. Early events in DSB repair involve loading of repressor complexes onto the chromatin via a PARP-dependent pathway. This leads to remodelling (demolition) of the pre-existing structure and the formation of a transiently compacted chromatin. This structure is then removed (transition) through a process linked to dePARylation, removal of H2A.Z and possibly through exchange of histone H3.3. This leads to the creation of repair-competent chromatin (relaxed chromatin) in which the H4 tail is acetylated, and 53BP1 and bromodomain proteins can be recruited to the break site. H3.3 may participate in several steps (indicated by ?) that still remain to be resolved.

Overall, it is clear that chromatin structure has a dynamic and evolving response during DSB repair. While the evidence is strong that repair is associated with a more open, accessible chromatin environment at the break, it is less clear how the surrounding regions may respond or how local chromatin organization alters as repair progresses. It is likely that changes in chromatin organization are dictated by the pre-existing chromatin organization, with compact regions such as heterochromatin experiencing both chromatin expansion and increased mobility to move the damaged region to the periphery of the heterochromatin for repair [70–73]. DSBs in regions that are already in an open conformation, such as genes or gene clusters, may require an initial compaction to block nearby transcription and retain overall chromatin integrity prior to repair. Processing of breaks by NHEJ or HR may require distinct types of chromatin reorganization, so that the damaged region may proceed through open and more compact conformations as repair progresses. Further, it is possible that the local chromosome territory, extending over megabase/chromosome regions, undergoes compaction, while the chromatin at the break (covering tens of kilobases) may be in a highly flexible, open conformation [95]. Understanding how chromatin responds to DSBs over these different scales of genomic organization, from whole chromosomes down to individual nucleosomes surrounding the DSB, remains an important area of study. Further, examining changes in the local chromatin organization at individual DSBs, rather than looking at ‘global averages’ across large cell populations, will help to unravel locus-specific changes in chromatin organization during repair.

5. H4Ac and the shift to (locally) open chromatin

As discussed above, chromatin reorganization during DSB repair is complex, involving both large-scale changes covering megabase domains as well as changes occurring directly at the DSB. Here, we will discuss how H4Ac on nucleosomes at the DSBs regulates both the local chromatin organization and the recruitment of key DSB repair proteins to the H4 tail.

A key event in the transition from the repressive state to the open structures needed for repair is acetylation of the H4 tail by the NuA4-Tip60 complex. H4Ac at DSBs is largely dependent on the TIP60 acetyltransferase (KAT5) subunit, which acetylates histone H4 at K5, K8, K12 and K16, as well as H2A at K5 and K15 [37,40,96,97]. NuA4-Tip60 is a 16 subunit remodelling complex containing multiple catalytic activities [1,98]. These include P400, a member of the INO80 family of SWI/SNF ATPases ([6,99]; the RUVBL1 and RUVBL2 AAA+ ATPases; and the TIP60 (KAT5) acetyltransferase. NuA4-Tip60 is rapidly recruited to DSBs and inactivation of subunits of the NuA4-Tip60 complex gives rise to multiple repair defects, including defects in processing damaged chromatin, increased genomic instability, loss of H4Ac and defective DSB repair [34,35,37–39,100–104]. The P400 subunit of NuA4-Tip60 is required for DNA repair in both yeast and mammalian cells [105–113]. P400 exchanges the histone variant H2A.Z onto nucleosomes at DSBs [34,66,107,114–117]. H2A.Z exchange has been shown to be important for repair of persistent DSB breaks [105], movement and tethering of damaged chromatin to the nuclear membrane [106–108], translesion synthesis [109] and overall DSB repair [106,110–113]. However, one key finding is that H2A.Z exchange at DSBs by P400 is required for the subsequent acetylation of the H4 tail by Tip60 during DSB repair [34,66], a process that is regulated by the acidic patch on the nucleosome surface.

As noted earlier, the acidic patch on the surface of the nucleosome is formed by acidic residues largely contributed by histone H2A and H2B [48]. These acidic amino acids form a shallow, acidic groove to which the unacetylated H4 tail can bind [43,52]. Acetylation of H4 (on lysine 16) blocks this interaction [1,43,48,52–54], preventing compact chromatin fibre formation and shifting nucleosomes to a more open, flexible structure. H2A.Z exchange at DSBs plays a critical role in altering the function of the acidic patch. H2A.Z has only weak (60%) homology to H2A/H2AX, but differs significantly in that it possesses an extended acidic domain compared with other H2A family members [48,118]. H2A.Z exchange onto nucleosomes therefore increases the charge density of the acidic patch, favouring the binding of the positively charged H4 tail to H2A.Z-nucleosomes [119,120]. This increased binding of the H4 tail promotes packing of nucleosome fibres [120,121] and limits nucleosome mobility. H2A.Z exchange at DSBs by the NuA4-Tip60 complex therefore promotes more compact chromatin and blocks acetylation of the H4 tail (figure 1) [34,66].

This suggests a complex, but co-ordinated series of nucleosome remodelling events carried out by the NuA4-Tip60 complex. Recruitment of NuA4-Tip60 leads to the rapid exchange of H2A.Z onto nucleosomes at the DSB by the P400 ATPase subunit. This creates a more dense acidic patch and favours binding of the H4 tail to adjacent nucleosomes and therefore reduces nucleosome mobility. The exchange of H2A.Z occurs immediately after damage, and corresponds to the period during which there is rapid, PARP-dependent accumulation of repressive proteins, including HDACs, at the DSB (figure 1; [89,94]). These HDACs may deacetylate H4, facilitating binding of the H4 tail to the acidic patch and contributing to the formation of a transient, compact chromatin conformation immediately after DNA damage. The initial accumulation of repressors and H2A.Z may therefore represent the initial reconstruction of the local chromatin, involving removal of pre-existing chromatin proteins and erasure of pre-existing histone modifications, such as acetylation. In this way, diverse chromatin domains can be rebuilt to create a common template for the DSB repair machinery. Demolition of the pre-existing chromatin structure is therefore the key initial step in DSB repair (figure 1). Further, this process creates an H4 tail that lacks acetylation and that is bound to the acidic patch on the surface of adjacent nucleosomes. Because many DNA repair proteins bind to the H4 tail, interaction of the H4 tail with the acidic patch provides a mechanism for regulating binding of key factors, such as 53BP1, during the initial processing of the DSB.

6. Nucleosome disassembly and H4 acetylation at double-strand breaks

The acetylation of H4 is largely driven by disruption of the interaction between the unacetylated H4 tail and the acidic patch on the nucleosome surface. This involves both removal of H2A.Z and, potentially, exchange of the histone variant H3.3.

(a). H2A.Z removal

The H2A.Z is only retained at DSBs transiently, with H2A.Z being rapidly removed from the damaged chromatin by the histone chaperone ANP32E (figure 1) [66,117,122]. ANP32E removes the entire H2A.Z-H2B dimer [122,123], a process potentially aided by the INO80 remodelling ATPase [6,99,117,124,125], creating partially disassembled nucleosomes lacking any acidic patch. This releases the H4 tail, allowing it to be acetylated by the Tip60 subunit of the NuA4-Tip60 complex. This process occurs with a similar time course to the loss of chromatin PARylation and release of various repressors from the chromatin (figure 1). In this way, dynamic exchange of H2A.Z can be used to control both chromatin compaction and H4Ac during the earliest moments of DSB repair. Further, this removal of H2A.Z, coupled with loss of PARylation, provides a regulated transition from a more compact, repressive chromatin structure to the more open, acetylated chromatin structure required for DSB repair (figure 1).

(b). H3.3: a new partner in double-strand break repair?

In addition to H2A.Z exchange, a second histone variant, histone H3.3, is exchanged onto chromatin during DSB repair [74,126–128]. H3.3 is expressed throughout the cell cycle and is deposited in regions that are transiently free of nucleosomes [129–131], such as genes [27,132–134]. H3.3 is enriched in telomeric and pericentromeric heterochromatin [135,136]. This pattern of H3.3 localization is maintained by two H3.3 chaperones—HIRA directs H3.3 to genes, whereas Atrx/Daxx maintains H3.3 in heterochromatin [130,137]. H3.1/H3.2 and H3.3 only differ by five amino acids, with these differences largely restricted to amino acids 87–90, the region that interacts with the H3.3-specific chaperones HIRA and ATRX/DAXX [27]. Because H3.3 and H3.1/H3.2 have virtually identical sequence, H3.3 is unlikely to alter the overall structure of the nucleosome. However, H3.3 has been associated with loss of histone H1 and a reduction in higher-order chromatin folding [132,138,139]. Interestingly, nucleosomes containing both H2A.Z and H3.3 are highly unstable and easily disassembled [140,141], suggesting that H3.3 and H2A.Z work together to create open chromatin structures during DSB repair.

H3.3 is important for maintaining genome stability. H3.3 depletion leads to mitotic defects, including anaphase bridges and increased DNA damage [126,142]. H3.3 exchange is required to restore transcription [128] and promote replication fork progression following UV damage [127]. H3.3 is exchanged onto the chromatin at DSBs by the CHD2 remodeller [74,143] where it promotes a rapid chromatin expansion and loading of NHEJ repair proteins. Further, H3.3 is deposited onto nucleosome-free regions that are created during processing of DSBs [144]. H3.3 exchange is a rapid, early event dependent on PARP (figure 1) [74], and therefore occurs during the initial formation of repressive chromatin associated with H2A.Z exchange [34,66]. Because nucleosomes containing H3.3 and H2A.Z are unstable [140,141], H3.3 exchange may destabilize H2AZ-nucleosomes, facilitating H2A.Z removal (by ANP32E) and promoting both H4Ac and DSB repair. Further, H3.3 is deposited onto nucleosome-free regions, such as transcribed genes [27,132–134] but is also important for chromatin reassembly following repair [127,144]. H3.3 deposition may therefore occur throughout repair, promoting the formation of open chromatin domains, as well as being essential for reassembly of chromatin following repair. How sustained H3.3 deposition can be maintained during DSB repair is not yet clear, but a recent report provides some insight. Pradhan et al. [145] have shown that P400 (a component of the NuA4-Tip60 complex) can catalyse exchange of H3.3 onto nucleosomes during transcription. This raises the possibility that NuA4-Tip60 may exchange both H2A.Z and H3.3 onto nucleosomes at DSBs. Previous work has shown that the CHD1 and CHD2 remodellers also function as H3.3 exchangers during transcription [146,147] and DSB repair [74]. This suggests that CHD2 may be responsible for the initial accumulation of H3.3 immediately after DNA damage (figure 1), while NuA4-Tip60 (which is retained at DSBs for an extended period; [35,102]) may function to maintain H3.3 throughout the repair process, and may be important for depositing H3.3 onto nucleosome-free DNA following successful DSB repair. Probing the mechanistic and temporal links between CHD2, NuA4-Tip60 and the exchange of H2A.Z and H3.3 onto damaged chromatin remains an outstanding issue. Further, there are several somatic mutations in H3.3 identified in paediatric cancers which can potentially disrupt H3.3 methylation and function [27,148]. Understanding how exchange of tumour-derived H3.3 mutations at DSBs influences genomic stability and tumorigenesis will be important.

(c). A model

Overall, we propose that the initial formation of repressive-like chromatin during the first few minutes after DNA damage serves to remodel the local chromatin, removing local H4 acetylation and promoting binding of the H4 tail to the acidic patch (figure 1). This initial ‘demolition’ of the original chromatin is aided by the recruitment of HDACs and H3K9 methyltransferases, and exchange of H2A.Z. H2A.Z exchange also increases the charge density of the acidic patch, increasing binding of the unacetylated H4 tail and promoting interaction between adjacent nucleosomes. Next, there is a ‘transition’ period in which local nucleosome remodelling, including dePARylation (through PARG activity), occurs, leading to the release of repressor complexes. This process also requires H2A.Z removal (by ANP32E), which eliminates the acidic patch and releases the H4 tail. This process is probably aided by exchange of H3.3, which can destabilize H2A.Z-nucleosomes and may facilitate H2A.Z removal by ANP32E. The release of the unmodified H4 tail from the nucleosome surface now allows it to be acetylated by Tip60. This acetylation further reduces inter-nucleosomal interactions and promotes the shift to a more open chromatin organization. Further, this process releases the H4 tail in an acetylated form where it can function both to recruit bromodomain proteins and to regulate recruitment of proteins such as 53BP1. This model provides an overall guide to how chromatin is reorganized during DSB repair. However, it is important to note that there may be significant mechanistic differences when comparing, for example, DSBs in heterochromatin, which are dependent on KAP-1 phosphorylation [62,65,67], with more open structures, including transcribed genes. It is, therefore, likely that there will be significant variation in the dynamics of chromatin organization during the initial detection and processing of DSBs in different chromatin domains. Exploring the dynamics of chromatin reorganization at individual DSBs will be needed to address this issue.

(d). Acidic patch and DNA repair: a quick aside

The above model indicates that the acidic patch is an important player in DSB repair, and several studies provide support for this idea. For example, the acidic patch is important for ubiquitination of the chromatin and H2A by the RNF8 and RNF168 ubiquitin-ligases [18,19,149,150]. Although RNF168 does not bind to the acidic patch, the charged surface allows RNF168 to ubiquitinate K13/K15 of H2A [150]. Further, in the absence of H4 acetylation, the H4 tail remains bound to the acidic patch [66] and both chromatin ubiquitination and 53BP1 loading are blocked [35,39,77,151]. Nucleosome remodelling through H2A.Z exchange and H4 acetylation [66] may therefore be essential for H2A ubiquitination and recruitment of 53BP1 [18,150]. It is also possible that exchange of H2A or H2AUb may provide a mechanism to control overall levels of histone ubiquitination at the DSB. The acidic patch and the nucleosome surface may therefore create a flexible binding domain that functions as a hub for regulating damage-mediated histone modifications, as well as providing a docking site for many of the chromatin modifiers involved in DSB repair.

7. The H4 tail and DNA repair

Although the previous discussion has focused on how acetylation of the H4 tail can alter histone–DNA interactions and disrupt interactions between neighbouring nucleosomes, the H4 tail also contains multiple modifications that can recruit specific DSB repair proteins to the chromatin. These include methylation of lysine 20 (H4K20me2), which is required for 53BP1 binding, and acetylation of lysines 5, 8, 12 and 16, which function together to recruit proteins containing bromodomains to sites of damage. Here, we will briefly review how specific methylation and acetylation signatures at DSBs regulate loading of DSB repair proteins.

(a). H4K20me2 and 53BP1

53BP1 is recruited to nucleosomes at DSBs through binding of its ubiquitin interacting motif (UIM) to H2A ubiquitinated at lysines 13 and 15 (H2AK13/15) [19,149] and its tudor domain to H4K20me2 [152]. H2A ubiquitination by RNF168 is increased after DNA damage [19,149], whereas 53BP1 relies extensively on pre-existing H4K20me2 for binding to DSBs. However, H4K20me2 can promote formation of heterochromatin, with the unacetylated H4K20me2 tail promoting chromatin compaction [153]. As noted in the previous section, the H4 tail is directly associated with the acidic patch on the nucleosome surface immediately after DNA damage, raising issues of how 53BP1 may be able to access H4K20me2 if the H4 tail remains associated with the nucleosome surface. Clearly, processes such as ubiquitination of histones [18,150,154], dynamic exchange of H2A.Z [34,66,114,116,117] and H3.3 [74], and H4 acetylation, which destabilize nucleosomes during DSB repair, can mediate the release of the H4 tail from the nucleosome surface. This, coupled with H2A ubiquitination, would provide both H2AUb and H4K20me2 for 53BP1 binding.

However, cells contain many proteins that compete for binding to H4K20me2 [155], and two of these, the L3MBTL1 tumour suppressor [156–159] and the lysine demethylase KDM4A [160], can block loading of 53BP1. This indicates that there is significant competition for binding to the H4 tail during remodelling events that occur during DSB repair. Consequently, both L3MBTL1 and KDM4A must be actively removed during DSB repair to allow efficient 53BP1 binding. Removal of L3MBTL1 and KDM4A involves ubiquitin-dependent pathways requiring proteasomal degradation of KDM4A [160], and removal of L3MBTL1 by the VCP/p97 ATPase complex [154,156,161], both of which are specifically activated by DSB repair. This implies that processing of the unacetylated H4 tail and its release from the nucleosomal surface during DSB repair must be tightly regulated. Exposure of the unacetylated H4K20me2 tail may lead to competition between L3MBTL1, KDM4A and 53BP1 for binding to the H4 tail. Ubiquitination and removal of L3MBTL1 and KDM4A from H4K20me2 by ubiquitin-ligases is therefore critical for loading 53BP1, and underscores the importance of ubiquitin in regulating 53BP1 recruitment and DSB repair [11,18,19,149,150,154,162,163]. Further, it indicates that the cell must exercise precise control of binding partners for the H4 tail during DSB repair by tightly regulating access to H4K20me2.

(b). H4Ac and 53BP1

A second level of regulation of the H4 tail is provided by direct acetylation of H4 at four key lysines—H4K5/K8/K12/K16. H4Ac is largely carried out by Tip60, as part of the NuA4-Tip60 complex, during DSB repair [39,97,100,102,104,164]. Tip60 is essential to promote genome stability and favours HR. Because 53BP1 functions to limit end-resection and blocks HR [165–169], this implies that Tip60 opposes 53BP1 binding. In fact, acetylation of H4K16 by Tip60 blocks binding of 53BP1 to the adjacent H4K20me2 [77,170]. Further, the Tip60 complex can itself bind to H4K20me2 (through the MBTD1 complex) and limit ubiquitination of H2A by directly acetylating the H2A ubiquitin site [37]. Although the Tip60 complex itself can bind to H4K20me2 [37], whether this interaction is limited by H4K16Ac is not clear. Binding of the Tip60 complex to H4K20me2, coupled with loss of H2AUb (which is required for 53BP1 binding [19]), together can prevent 53BP1 binding and promote HR [37]. These results suggest a highly dynamic set of H4 modifications occurring during DSB repair. Acetylation of H4K16 and H2AK15 by the Tip60 complex can block ubiquitination of H2A, preventing interaction of 53BP1's UIM with H2AUb and the tudor domain with H4K20me2. This inhibits 53BP1 loading at DSBs and favours HR-mediated repair over NHEJ [37,77,166–168]. Dynamic changes in H4Ac can therefore have a profound influence on the choice of DSB repair pathway.

(c). H4Ac and bromodomains

A third level of regulation of the H4 tail occurs at the level of binding to acetylated lysine residues in the H4 tail. Bromodomains are small hydrophobic pockets that can bind to acetylated lysine in a sequence-specific manner [171]. There are at least 61 bromodomains in the human genome which are found in proteins involved in transcription, chromatin modification and DNA repair. Bromodomain-containing proteins involved in DNA repair include BRD4 [61], ZMYND8 [172], BRD8 (a subunit of the NuA4-Tip60 complex [98]), TRIM28 (KAP-1 [64,67]), ACF1 [59,62] and the ATPase subunits of SWI/SNF remodelling complexes (including BRG1) [56,173,174]. Neither BRD8 nor TRIM28 interacts directly with acetylated H4 residues. TRIM28 (KAP1) is phosphorylated by the ATM kinase during DSB repair [64], and this phosphorylation is required for chromatin relaxation and DSB repair within heterochromatin [67,68]. However, while TRIM28 contains tandem PHD-BRD domains, TRIM28 binding is not regulated through interactions with acetylated lysine [175]. Instead, the PHD-BRD structure in TRIM28 may function as a SUMO-ligase [176]. Similarly, the BRD of ACF1 does not play a major role in recruiting this protein to DSBs, with ACF1 loading probably mediated through the PHD domains [62]. The ATPase subunits of the SWI/SNF nucleosome remodelling complexes, BRG1 and BRM, also play a critical role in DSB repair [174,177]. BRG1 is recruited to damage sites through binding of its bromodomain to acetylated H3 [173,178], although other subunits of SWI/SNF contribute to this interaction [174]. Therefore, although many proteins required for DSB repair contain BRD domains, these BRDs are not always essential or required for their recruitment to DSBs.

The BRD domains of two proteins, BRD4 [61] and ZMYND8 [172], have been implicated in DSB repair. BRD4 contains two tandemly arrayed bromodomains with specificity for H4K8Ac/K12Ac [179,180]. BRD4 is important for transcriptional elongation and recruiting transcriptional co-activators to genes and is a key target for bromodomain inhibitors in cancer therapy [171,181,182]. Loss of BRD4 leads to genome-wide relaxation of chromatin structure and the rapid expansion of γH2AX domains after DNA damage. The presence of BRD4 on H4Ac domains may serve to limit the spreading of γH2AX and other modifications, ‘insulating’ the surrounding chromatin from DNA damage signalling, and provide a mechanism for controlling the spreading of damage-induced chromatin modification. Although BRD4 can bind directly to H4Ac [179,180], BRD4 does not accumulate at DSBs [61]. This lack of BRD4 accumulation at DSBs means that the increased H4Ac at DSBs does not serve as a binding site for BRD4. Instead, it is the pre-existing distribution of BRD4 along the chromatin, rather than damage-induced changes in BRD4 localization, that is critical for BRD4's function during DSB repair. BRD4 is not, therefore, directly recruited to nascent H4Ac generated at DSBs.

Recent work has suggested that another bromodomain protein, ZMYND8, is actively recruited to H4Ac at DSBs [172,183]. ZMYND8 is a transcriptional repressor containing a bromodomain, PWWP and PHD chromatin-binding domains [184]. These domains are tandemly arranged (PHD-BRD-PWWP) so that ZMYND8 functions as a combinatorial reader of multiple histone marks, including H3K14Ac/H3K4me [183,184]. ZMYND8 is recruited to DSBs through direct interaction with H4K8Ac/K12Ac, a process dependent on Tip60 (for H4Ac) and the bromodomain of ZMYND8 [172]. ZMYND8 recruitment is essential to repress local transcription and promote HR-directed repair. Further, ZMYND8 retention at DSBs may require all three (PHD-BRD-PWWP) domains for maximal binding [58,172,183], suggesting that ZMYND8 is reading both H4Ac and H3 methylation at DSBs. Importantly, ZMYND8 serves as a platform to recruit the NuRD complex [172], a remodelling complex containing CHD3/CHD4 remodellers and HDAC activity, which is important for DSB repair [7,57,75]. This will bring HDAC activity to the H4 tail and may facilitate local editing of acetylated H4 and potentially other acetylated histones such as H3. For example, ZMYND8/HDAC complexes bound to H4K8Ac/K12Ac may allow deacetylation of e.g. H4K16Ac, and thereby facilitate 53BP1 binding [37,77,170]. Such results would be consistent with previous studies showing that loss of HDACs leads to defects in NHEJ repair [87], conditions that would be expected to block 53BP1 binding.

The wide range of binding partners for the H4 tail, and the presence of both histone methylation and acetylation marks, indicate that the H4 tail is a key contributor to the repair of DSBs. As shown in figure 1, precise regulation of H4 acetylation and methylation signatures during DSB repair drives DSB repair. H4Ac can function to modulate histone–DNA and inter-nucleosomal interactions, creating relaxed chromatin structure (figure 1). However, H4Ac also plays an active role in recruiting binding partners, such as ZMYND8, and in limiting interaction of the acetylated H4 tail with the acidic patch on adjacent nucleosomes. H4Ac is also critical for limiting binding of 53BP1 and promoting HR. There are also key roles for H4 acetylation in regulating the binding of repressors such as L3MBTL1 and KDM4A to H4K20me2. Understanding the dynamic changes in binding of these proteins to the H4 tail during DSB repair, and elucidating how they contribute to 53BP1 loading and other processes, are ongoing issues.

8. Conclusion

The close co-operation between the chromatin remodelling machinery and DSB repair proteins is vital to maintaining genome stability. This involves an initial loading of multiple repressive complexes and processing of the chromatin. This creates a transient repressive state in which the H4 tail is bound to the acidic patch on adjacent nucleosomes. Although this first step in the DNA damage response is transient, lasting only a few minutes, it is likely to be important for repressing ongoing transcriptional activity in the region, rewriting the local histone code, sequestering the damaged chromatin and setting the stage for subsequent DNA repair steps. By immobilizing the H4 tail on the acidic patch, the cell can also control both the acetylation status and binding partners of the H4 tail during this initial stage of DSB repair. In this way, the pre-existing chromatin environment can be rapidly converted to a format appropriate for DSB repair.

Transition to open chromatin involves disassembly of the nucleosome and loss of the acidic patch, releasing the unacetylated H4 tail from the nucleosome surface. The acidic patch on the nucleosome surface, and binding of H4 to it, are therefore critical for DSB repair. The newly accessible H4 tail now becomes amenable to modifications by histone-modifying complexes, including the Tip60 complex. The H4 tail can recruit 53BP1 (via H4K20me2), or repressors such as L3MBTL1, which can compete for 53BP1 binding. Further, acetylation of the H4 tail can itself block binding of 53BP1, as well as limiting the ability of H4 to rebind to the acidic patch. Finally, H4Ac can provide binding sites for readers of acetylation (bromodomains), allowing regulated recruitment of proteins such as ZMYND8 and BRD4 which are important for DSB repair. The H4 tail and its extensive set of modifications and binding partners are therefore a critical hub for directing the DSB repair machinery. Understanding how specific H4Ac codes and modifications are written and erased during DSB repair, and how each modification contributes to nucleosome function or recruitment of specific proteins, will provide insight into the importance of the H4 tail for DSB repair.

Authors' contributions

S.D., O.G.-Y. and R.M. and B.D.P. contributed to ideas and analysis. B.D.P. and S.D. wrote the manuscript.

Competing interests

The authors have no competing interests to report.

Funding

This work was supported by NIH grant no. CA64585, CA93602 and CA177884 to B.D.P.