|Home | About | Journals | Submit | Contact Us | Français|
MOF (MYST1) is the major enzyme to catalyze acetylation of histone H4 lysine 16 (K16) and is highly conserved through evolution. Using a conditional knockout mouse model and the derived mouse embryonic fibroblast cell lines, we showed that loss of Mof led to a global reduction of H4 K16 acetylation, severe G2/M cell cycle arrest, massive chromosome aberration, and defects in ionizing radiation-induced DNA damage repair. We further showed that although early DNA damage sensing and signaling by ATM were normal in Mof-null cells, the recruitment of repair mediator protein Mdc1 and its downstream signaling proteins 53bp1 and Brca1 to DNA damage foci was completely abolished. Mechanistic studies suggested that Mof-mediated H4 K16 acetylation and an intact acidic pocket on H2A.X were essential for the recruitment of Mdc1. Removal of Mof and its associated proteins phenocopied a charge-neutralizing mutant of H2A.X. Given the well-characterized H4-H2A trans interactions in regulating higher-order chromatin structure, our study revealed a novel chromatin-based mechanism that regulates the DNA damage repair process.
In eukaryotes, DNA is packaged with core histones and other nonhistone chromosomal proteins into several orders of chromatin structure with increasing compaction. Many cellular processes, including transcription, DNA replication, and DNA damage repair (DDR), are regulated in the context of chromatin. Recent studies have shown that histone modification (e.g., RNF8 and RNF168) and chromatin-remodeling activities (e.g., INO80 and SWR1) facilitate the accumulation and function of DNA repair proteins at the damage foci (52). Most of the regulations are achieved at the level of nucleosomes. Specifically, chromatin regulatory activities can either alter nucleosome structure and location or modulate histone-DNA contacts to promote association of trans-acting factors with DNA and recruit important components of the signaling cascade to DNA damage repair centers (16). Among them, the function of Saccharomyces Cerevisiae histone acetyltransferase (HAT) NuA4 and its mammalian homolog, Tip60, have been well characterized in this process (45, 52). Mutations in NuA4 or its lysine substrates on histone H4 tail in yeast led to increased sensitivity to DNA-damaging reagents and impaired double-strand repair by nonhomologous end joining (NHEJ) (3, 11). In higher eukaryotes, it has been shown that Tip60 regulates DNA repair through acetylation of both H2A and H2A.X, which facilitates polyubiquitination and dynamic exchange of H2A.X at the damage foci, and of histone H4 at lysine 5 (H4 K5), H4 K8, and H4 K12, which facilitates nucleosome remodeling and establishes less condensed nucleosome arrays (42, 44). Although regulation of DNA damage repair by higher-order chromatin structures has been proposed in these studies, mechanistic details remain unclear.
Unlike most histone modifications, H4 K16 acetylation (H4 K16ac) is unique for regulating higher-order chromatin structures beyond the level of nucleosomes. It was first reported in the crystal structure of nucleosome particles by Luger et al. that a basic region in the H4 tail (amino acids 14 to 19) is involved in the internucleosome interaction with an acidic pocket on the H2A-H2B surface in the crystal lattice (28). Subsequent studies showed that this H4-H2A interaction is essential for folding of nucleosome arrays into secondary and tertiary chromatin structures (5). The conclave acidic surface on H2A (and most H2A variants) is formed by several highly conserved negatively charged residues. Neutralizing these negative charges by mutations or by binding of a herpesvirus-derived latency-associated nuclear antigen (LANA) peptide promotes salt-induced chromatin condensation in vitro and apparent heterochromatination in vivo (6, 55). Acetylation of K16, which neutralizes the positive charge on the basic patch of H4, weakens the interaction of the H4 tail with the acidic pocket on H2A and reduces the propensity of nucleosome arrays to self-associate into the 30-nm chromatin fiber (21, 37, 40). In addition to this well-characterized function, H4 K16 acetylation was also shown to block long-range internucleosome interactions imposed by linker histones (21). Despite extensive biochemical and biophysical studies of H4 K16 acetylation in higher-order chromatin structures in vitro, its physiological implications in the control of important nuclear processes in vivo remain unclear.
The major enzyme that acetylates histone H4 K16 in mammals is MOF (also called MYST1 or KAT8) (25, 33). It is a highly conserved MYST family HAT, sharing the same domain structure and sequence homology with Tip60 (51, 53). Unlike most HATs that targets multiple sites on histones, MOF activity on nucleosomes is restricted to K16 on the histone H4 tail and is tightly regulated in two distinct complexes: the MOF-MSL complex and the MOF-MSL1v1 complex (reviewed in reference 25). In addition to H4 K16, the MOF-MSL1v1 complex, but not the MOF-MSL complex, also acetylates nonhistone substrates such as p53 and plays important roles in transcription activation (26). Despite extensive biochemical characterization (10, 26, 41, 48), the in vivo function of Mof in higher eukaryotes was not well understood. In vivo studies using mouse models showed that Mof is essential for vertebrate development and that constitutive ablation of Mof leads to peri-implantation lethality in mouse embryos (14, 49). In one study, Mof-null mice showed a marked delay in developmental progression with proliferation arrest and death at about E7.5 (14). In the second study, all Mof−/− embryos failed to develop beyond the expanded blastocyst stage and died at about the implantation stage (E4.5) (49). In both studies, Mof-null embryos showed massive abnormal chromatin aggregations (14, 49), suggesting a crucial role for Mof in the maintenance of chromatin structures. However, the early lethality of Mof-knockout mice prevented detailed loss-of-function studies for Mof in a defined genetic background in higher eukaryotes.
For in-depth study of the functions of Mof and histone H4 K16ac in the DNA damage response, we have established a conditional Mof-knockout mouse model. In this model, the floxed Mof allele can be deleted by 4-hydroxytamoxifen (4-OHT)-induced expression of Cre recombinase. Using the derived Mof+/+ Cre-positive (Cre+) and Mof−/− Cre+ mouse embryonic fibroblast (MEF) cell lines, we found that Mof is essential for global H4 K16 acetylation, cell proliferation, and maintenance of genome stability. Importantly, Mof was required for efficient repair of DNA damage induced by ionizing radiation (IR). Unlike Tip60, Mof did not affect ATM activation or γH2A.X accumulation at the damage foci. Instead, it is crucial for the recruitment of repair mediator proteins such as Mdc1 (mediator of DNA damage checkpoint 1), 53bp1 (p53 binding protein 1), and Brca1 (breast cancer type 1 susceptibility protein). Further mechanistic studies revealed that Mof and H4 K16 acetylation probably regulate the DNA repair process by modulating interactions of the H4 tail with the acidic pocket of H2A.X. Loss of H4 K16 acetylation or neutralization of the charge of acidic pocket on H2A.X abolished Mdc1 binding at the DNA damage foci. Together, our results demonstrated a novel mechanism for the chromatin-mediated DNA damage repair process.
To generate conditional Mof-knockout mice, a targeting vector was constructed (see Fig. Fig.1).1). This vector was introduced into R1 embryonic stem cells via electroporation. Three homologous recombinants were obtained. The embryonic stem cells that carry Mofflox/+ alleles were injected into C57BL/6 blastocysts. The chimeric mice were backcrossed with C57BL/6 to get F1 mice (Mofflox/+). Mofflox/+ mice were backcrossed six times onto the C57BL/6 background. Homozygotes were generated by crossing heterozygote Mofflox/+ mice with CAGG Cre-ER mice. The double heterozygotes were crossed to generate Mofflox/flox CAGG Cre-ER mice. All animal experiments were carried out according to the guidelines for the care and use of laboratory animals of the University of Michigan. The mice used in this study were genotyped by PCR analysis. The primers and their sequences were as follows: primer Mof-F, AACGTGGCATTGATTATTG; primer Mof-R, GCTTCTTTCTCATCCTGGTAG; primer Cre-s, GCCAGCTAAACATGCTTCATC; and primer Cre-a, ATTGCCCCTGTTTCACTATCC. The multiplex PCR produces a 150-bp band for the wild-type (WT) allele, a 200-bp band for the floxed allele, and both bands for heterozygous mice; and the Cre primer produces a 727-bp band.
Reverse transcription-PCRs (RT-PCRs) were performed to detect truncated Mof transcripts after 4-OHT treatment. The primers are (i) AAGAACCGACTCGCACTGAC (sense) and (ii) AAGACAAAGGGCTCCACATC (antisense). The wild-type Mof transcript produces a 522-bp band, and the Mof-knockout transcript produces a 219-bp band.
The antibodies used in this study include anti-H3 Ser10phos (Cell Signaling), anti-acetyl-histone H4 K16 (07-329; Millipore), anti-ubiquityl-histone H2A clone E6C5 (05-678; Millipore), anti-p19Arf (M156; Millipore), anti-histone H3 (Millipore), anti-γH2AX (active motif, 39117), anti-H2A (active motif, 39209), anti-histone H4 dimethyl Lys20 (active motif, 39539), anti-acetyl-histone H4 K12 (active motif, 39165), anti-acetyl-histone H4 K5 (active motif, 39583), anti-acetyl-histone H4 K8 (active motif, 39171), anti-ATM Ser1981p clone 10H11.E.12 (active motif, 39529), anti-p16Ink4a (abA300-992A; abcam), anti-MOF (MYST1) (BL2118; Bethyl Labs), anti-TIP60 (N-17; Santa Cruz), anti-p21(C-19; Santa Cruz), anti-β-actin (Cell Signaling), anti-p53S15p (Cell Signaling), anti-Smc1S957p (Cell Signaling), anti-Chk2 clone 7 (Cell Signaling), and anti-Chk1S345p (Cell Signaling). Anti-Brca1, anti-53bp1, anti-Mdc1, and anti-γH2AX were described previously (18a).
MEFs were isolated from E13.5 embryos and grown under standard culture conditions. The Mofflox/flox CAGG Cre-ER MEF cells were cultured in the presence of 200 nM 4-OHT for 4 days to delete Mof before each experiment. The population doubling time was measured using the 3T9 protocol described previously (50). Cells from passage 7 (p7) of wild-type and Mof-knockout MEFs were used to obtain growth curves. For genome instability studies, metaphase chromosomes were prepared as described previously. The glass slides were then stained with 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen). Images were acquired on an Olympus BX-61 microscope and viewed with SKYview software (ASI).
Passage 8 MEFs were evaluated for cellular senescence using a senescence-associated-β-galactosidase (SA-β-Gal) staining kit (senescence cell staining kit; Sigma). SA-β-galactosidase staining was performed in the 24-well plate format, according to the manufacturer's instructions. Senescent cells are stained blue.
Primary MEFs (passages 5 and 6) were harvested and stained with propidium iodide (PI; 10 μg/ml in 0.5% bovine serum albumin [BSA] in 1× phosphate-buffered saline [PBS]) for 30 min at 37°C. The samples were run on a BD LSRII flow cytometer and analyzed using the BD FACS Diva program (version 6.0) and ModFit32 program. To identify cells in M phase, MEFs were fixed in 70% ethanol (following 1 h of recovery after exposure to 5 Gy IR or to no IR as a control) and stained using anti-phosphorylated H3 Ser10 (Cell Signaling). After incubation with secondary anti-rabbit IgG-AlexaFluor488 (1:200 in 0.5% BSA in PBS, 1 h at room temperature), cells were stained with 50 μg/ml PI. Flow cytometry was performed using a BD LSRII flow cytometer and were analyzed using the BD FACS Diva program (version 6.0) and the WinList3 (version 2) program.
MEFs were grown on coverslips (2 × 104 cells per well in a 24-well dish) and fixed in 4% paraformaldehyde for 10 min. Fixed cells were washed with PBS and permeabilized in 0.5% Triton X-100 for 5 min. After 1 h of blocking with 5% fetal bovine serum (FBS) in PBS, cells were stained with primary antibodies diluted in PBS (0.05% FBS) for 2 h at room temperature and then washed and incubated in secondary antibody conjugated with fluorophores. Cells were stained with DAPI, mounted on glass slides, and analyzed using a fluorescent microscope (Zeiss).
Strand break repair was analyzed by single-cell agarose gel electrophoresis under neutral conditions, as described previously (54). Cells were irradiated (80 Gy) and harvested immediately or at 1 h or 6 h postexposure. The images were analyzed for comet tail moment using CometScore software. At least 40 images were processed for each time point for each cell line.
The dividing line between green fluorescent protein (GFP; homologous recombination)-positive and -negative cells was set to the background level for the GFP-positive cells in the internal control (not transfected with I-SceI). This gate was then applied to the I-SceI-positive counterparts to determine homologous recombination efficiency.
Transfection was performed in 60-mm dishes, using Lipofectamine 2000 (Invitrogen) transfection for Mofflox/flox CAGG Cre-ER-positive MEF cells, according to the manufacturer's instruction. The short hairpin RNA (shRNA) for Msl1v1 and small interfering RNA (siRNA) for Msl1 were used to knock down the expression of Msl1v1 and Msl1, respectively (sequences are available upon request). Mof−/− MEF cells transfected with the pLentilox RSV-Puro (UM vector core) vector, which expresses either FLAG-tagged WT or E350Q (mutant) Mof, followed by a 3-day selection with puromycin (200 ng/ml).
Acetyltransferase assays were performed as described previously (10). Recombinant proteins (~2 μg) were used. For all reactions, equivalent amounts of MOF-MSL1v1 protein (less than 100 ng) were used. Reactions were carried out at 30°C for 1 h in the presence of [3H]acetyl coenzyme A ([3H]acetyl-CoA).
In order to study the functions of Mof in vivo, we generated the conditional Mof alleles that can be inactivated through Cre/loxP-mediated deletion. The targeting vector was constructed to have a loxP site inserted between exons 3 and 4. A neomycin resistance gene selection cassette flanked by two loxP sites was introduced between exons 6 and 7 (Fig. (Fig.11 A). Thus, exons 4 to 6 will be deleted upon the expression of Cre recombinase (Fig. (Fig.1A).1A). This region encodes part of the catalytic MYST domain of Mof (amino acids 173 to 257) and is essential for the Mof acetyltransferase activity (data not shown). The Mofflox/+ mice obtained were then crossed with CAGG Cre-ER transgenic mice, which ubiquitously express Cre recombinase in the presence of 4-OHT (17). The heterozygotes for both Cre-ER and floxed Mof alleles were intercrossed to get Cre-ER-positive Mofflox/flox, Mofflox/+, and Mof+/+ mice (Fig. (Fig.1B).1B). Primary MEF cell lines were derived from E13.5 Mof+/+ Cre+ and Mofflox/flox Cre+ embryos to carry out the studies described herein.
Upon 4-OHT treatments, about 95% of the Mofflox/flox Cre+ cells have complete deletion of floxed Mof alleles. RNAs and proteins were extracted from these cells 4 days after 4-OHT treatment for further analyses. As shown in Fig. 1C, a shorter transcript of Mof was detected by RT-PCR in Mof−/− Cre+ cells, consistent with the removal of exons 4 to 6. Immunoblotting showed that the full-length Mof protein was absent in Mof−/− Cre+ cells (Fig. (Fig.1D).1D). Notably, we did not detect any truncated Mof protein using an antibody recognizing the first 50 amino acids of Mof (data not shown). It is likely that deletion of amino acids 173 to 257 of Mof destabilized the Mof protein in vivo. Consistent with previous reports that Mof is the major H4 K16 acetyltransferase in mammals (10, 41, 48), Mof deletion led to global reduction of H4 K16 acetylation in MEF cells, as demonstrated by both immunoblotting (Fig. (Fig.1D)1D) and immunofluorescence assay (Fig. (Fig.1E).1E). Expressing wild-type Mof in the Mof-knockout cells greatly enhanced H4 K16 acetylation (Fig. (Fig.1F).1F). In contrast, cells expressing a Mof E350Q mutant that is mostly defective in substrate binding (29) had significantly lower levels of H4 K16ac. The effect of Mof knockout on H4 K16ac is highly specific because acetylation of other lysine residues on the H4 tail was not affected (Fig. (Fig.1D).1D). Loss of Mof and H4 K16 acetylation did not affect the dimethylation of H4 K20, a nearby lysine residue in the same basic patch region of the H4 tail (Fig. (Fig.1D1D).
Characterization of the Mof−/− Cre+ cells revealed that Mof played important roles in cell proliferation. Using a 3T9 protocol for the cell proliferation assay (50), Mof−/− Cre+ cells exhibited decreased rates of population doublings after p5 compared to the rate for wild-type (Mof+/+ Cre-negative [Cre−]) primary cells (Fig. (Fig.22 A, left). Cell growth almost completely stopped by p7 for Mof−/− Cre+ cells (Fig. (Fig.2A,2A, right). The slow growth of Mof−/− Cre+ cells was not a result of 4-OHT treatment, since Mof+/+ Cre+ cells show population doubling and growth similar to those of Mof+/+ Cre− cells (Fig. (Fig.2A).2A). The proliferation defects of Mof−/− Cre+ cells can be completely rescued by wild-type Mof and only partially by Mof E350Q (Fig. (Fig.2B).2B). The partial rescue by Mof E350Q was probably due to the residual acetyltransferase activity of this mutant (Fig. (Fig.1F).1F). Furthermore, more than 50% of p7 Mof−/− Cre+ cells stained positive for SA-β-Gal, suggesting premature senescence (Fig. (Fig.2C,2C, bottom). In contrast, very few SA-β-Gal stain-positive cells were found at this stage in control Mof+/+ Cre+ cells (Fig. (Fig.2C,2C, top). Consistent with the SA-β-Gal staining results, we found dramatic upregulation of several senescence markers, such as p16Ink4a, p19Arf, and p21 (7), in Mof−/− Cre+ cells (Fig. (Fig.2D).2D). An increase in cell death for Mof−/− Cre+ cells compared to that for wild-type cells was also observed in late passages.
We next examined the genome stability of primary Mof−/− Cre+ cells by karyotype analysis. The p6 Mof+/+ Cre+ and Mofflox/flox Cre+ MEF cells were treated with 4-OHT to delete Mof. Metaphase chromosomes from two independent Cre-positive Mof+/+ or Mof−/− cell lines were prepared and analyzed for chromosome anomalies. Although mitotic chromosome condensation was normal, we found an increase in aneuploidy and polyploidy (4N) as well as higher frequencies of chromosome breaks or fragmentations in Mof−/− cells (Fig. (Fig.2E,2E, indicated by arrows). As summarized in Fig. Fig.2F,2F, 54% of Mof−/− cells but only 12% of wild-type cells demonstrated aneuploidy or polyploidy by p6. Furthermore, 56% of Mof−/− cells but only 14% of wild-type cells contained chromosome breaks (Fig. (Fig.2F).2F). In most cases, multiple chromosome breaks or gaps were found in each metaphase. The chromosome anomalies increased dramatically with increasing numbers of cell divisions. Almost 20% more abnormal metaphases were observed in p6 Mof−/− cells than in p5 cells (data not shown). The accumulation of spontaneous genome instability in Mof−/− cells suggests that Mof plays important roles in maintaining genome integrity in higher eukaryotes.
The increase in chromosome anomalies in Mof−/− cells may result from defects in either cell cycle checkpoint or DNA repair. We first analyzed the cell cycle index of unsynchronized Mof+/+ and Mof−/− cell cultures. Fluorescence-activated cell sorter (FACS) analysis showed that a much higher percentage of Mof−/− cells was in G2/M phase (~47% versus 20% for Mof+/+ cells) (Fig. (Fig.3,3, top). In relation to this, there were fewer Mof−/− cells in G1 and S phases of the cell cycle (Fig. (Fig.3,3, top). When exposed to DNA-damaging agents such as 5 Gy IR, Mof−/− cells showed no IR-dependent changes in the cell cycle index (Fig. (Fig.3,3, bottom). Under similar conditions, wild-type MEF cells demonstrated IR-induced changes in both G2/M and S phases of the cell cycle: an almost 25% increase in the G2/M-phase population and a 6.5% decrease in the S-phase population (Fig. (Fig.3,3, bottom). Given the persistent G2/M-phase arrest both under normal culture conditions and under conditions of IR induction, it is likely that the genome instability observed in Mof-null cells is due to failure to repair damaged DNA.
To directly examine whether Mof−/− cells are defective in DNA repair, we first tested if repair by NHEJ is affected in Mof−/− cells after DNA damage. Using the neutral comet assay (54), both wild-type and Mof−/− cells were either mock treated (−IR) or exposed to 80 Gy IR (+IR) and harvested at various time points postexposure (Fig. (Fig.44 A). Although induction of DNA breaks was similar for wild-type and Mof-knockout cells, Mof−/− cells repaired the damaged DNA at a lower rate (Fig. (Fig.4A).4A). Compared to the results for wild-type cells, Mof−/− cells showed a modest decrease in repair efficiency (P < 0.0001), having a higher percentage of unrepaired DNA in the comet tail by 6 h (Fig. (Fig.4A,4A, bottom). Similar moderate defects in DNA repair were also observed for Mof−/− cells using a pulsed-field gel electrophoresis-based double-strand break (DSB) assay after IR (data not shown). In addition to NHEJ, we also examined the function of MOF in homology-directed repair (HR). To this end, we performed a gene conversion assay using a previously established U2OS cell line with integrated direct repeat-GFPs (DR-GFPs) (18). In this assay, we transfected cells with a replication-defective adenovirus expressing I-SceI together with either control siRNAs or MOF siRNAs. Expression of the endonuclease I-SceI in the cells leads to double-strand breaks in the DR-GFP reporter which are repaired by HR and which produce GFP-positive cells (36). Thus, the efficiency of HR repair is measured by flow cytometry for GFP-positive cells. As a negative control, cells transfected with adenovirus containing no I-SceI were analyzed in parallel. As shown in Fig. Fig.4B,4B, cells treated with MOF siRNA consistently produced 40% fewer GFP-positive cells than cells treated with control siRNAs (P < 0.0001). The knockdown efficiency for MOF siRNA in U2OS cells was ~50% (Fig. (Fig.4C).4C). Taken together, Mof is likely to play important roles in IR-induced DNA damage repair. During the revision of the manuscript, similar results were reported by Sharma et al. for 293 cells treated with MOF siRNAs (39).
To further study the potential involvement of Mof in DNA damage repair and the underlying mechanism, we decided to examine whether IR treatment leads to changes in Mof-mediated H4 K16 acetylation in wild-type MEF cells. Interestingly, we found that global H4 K16 acetylation was highly induced ~1 h after 10-Gy IR exposure (Fig. (Fig.55 A). Importantly, this induction was mostly specific for H4 K16 acetylation on the H4 tail, since no significant increase in the level of acetylation of other H4 lysine sites was detected after IR treatment (Fig. (Fig.5A,5A, right). Of note, the increase in the level of H4 K16ac occurred after H2A.X phosphorylation, which was upregulated 10 min after IR exposure (Fig. (Fig.5A).5A). We could not detect any damage focus-specific increase in H4 K16ac by immunofluorescence, probably due to the strong nuclear staining for H4 K16ac in cells prior to IR treatment (data not shown).
The increase in the level of H4 K16 acetylation at 1 h after IR suggests that Mof may function at a later stage relative to the times of ATM activation and H2A.X phosphorylation. To test whether Mof affects ATM signaling, we compared the level and kinetics of IR-induced ATM Ser1981 phosphorylation as well as γH2A.X in Mof+/+ and Mof−/− cells by immunoblotting. As shown in Fig. Fig.5B,5B, Mof knockout did not affect either the level or the kinetics of ATM autophosphorylation (Fig. (Fig.5B).5B). Furthermore, phosphorylation of important targets of the ATM signal transduction cascade, such as H2A.X, Chk1, Smc1, and p53 (23), was not affected in Mof−/− cells (Fig. (Fig.5C5C and data not shown). Of note, downregulation of γH2A.X was not affected by Mof deletion at later stages of the IR response. As shown in Fig. Fig.5C,5C, at 6 h after exposure to IR the level of γH2A.X in Mof−/− cells decreased to a level similar to that in wild-type cells. This observation is distinct from those for cells with null mutations of the closely related Tip60 (24) or the Tip60 complex component Trrap (32). Loss of function for either protein leads to delayed downregulation of γH2A.X at the DNA damage foci, an observation attributed to slower chromatin dynamics (12, 20, 24, 32). The fact that Mof and Tip60 have different effects on γH2A.X argues that they probably function at different steps in the DNA damage response.
We next investigated whether Mof regulates downstream events, such as the recruitment of repair mediator proteins such as Mdc1, 53bp1, and Brca1 to the DNA damage foci. Loading of these repair proteins to sites of DNA double-strand breaks serves to amplify DNA damage signaling and activate checkpoint controls (52). This process can be monitored by immunofluorescence for rapid formation of distinct IR-induced foci (IRIFs). To this end, we treated Mof+/+ and Mof−/− cells with 10 Gy IR and performed immunostaining with antibodies against γH2A.X, Mdc1, 53bp1, and Brca1 2 h after treatment. Consistent with the immunoblotting results, IR-induced γH2A.X focus formation was not affected in Mof−/− cells (Fig. (Fig.66 A). In contrast, IRIF formation for Mdc1, 53bp1, and Brca1 was mostly abolished in Mof−/− cells (Fig. (Fig.6A),6A), with more than 90% of cells staining negative for IRIFs (Fig. (Fig.6B).6B). Of note, Brca1 was redistributed in the nuclei of Mof−/− cells after IR, which is distinct from the distributions of Mdc1 and 53bp1 (Fig. (Fig.6A).6A). The mitigated recruitments for Mdc1, 53bp1, and Brca1 were not due to reduced expression of these genes (Fig. (Fig.6C)6C) or changed recruitment kinetics (data not shown). It is important to point out that the effect of Mof deletion on Mdc1 recruitment is distinct from the effects of other histone-modifying activities (e.g., those of Tip60, Rnf168, and Rnf8), which mostly affect steps downstream of Mdc1 binding (52). To confirm that Mdc1 binding to γH2A.X foci depends on Mof acetyltransferase activity, we transfected Mofflox/flox Cre+ cells with either wild-type Mof or Mof E350Q. The cells were then treated with 4-OHT to delete endogenous Mof. As shown in Fig. 6D and E, wild-type Mof significantly increased the numbers of H4 K16ac and Mdc1-positive cells in the population compared to those in the Mof mutant population (P < 0.0001), which had only a partial increase in the numbers of Mdc1 foci due to residual HAT activity (Fig. (Fig.6E).6E). The positively stained cells (~5%) in Mof−/− culture were probably due to incomplete deletion of endogenous Mof in these cells.
We previously demonstrated that Mof in mammalian cells resides in two distinct complexes: Mof-Msl and Mof-Msl1v1 (10, 26). While both Mof complexes are able to acetylate nucleosomal H4 K16, Mof-Msl1v1 has an expanded substrate spectrum and is capable of acetylating nonhistone substrates, such as p53 (4, 26). To test which Mof complex is important for Mdc1 recruitment and to distinguish the contributions of histone versus nonhistone acetylation, we examined Mdc1 IRIF formation in MEF cells treated with either Msl1 siRNAs or Msl1v1 shRNAs. As shown in Fig. Fig.7,7, about 30% of the MEF cells showed reduced H4 K16ac after siRNA and shRNA treatment for Msl1 and Msl1v1, respectively. The reduction of H4 K16ac was closely correlated with disappearance of the Mdc1 IRIF after DNA damage in both knockdown cells (Fig. 7B and C). In fact, there was no significant difference between Msl1 and Msl1v1 siRNA/shRNA treatments (Fig. (Fig.7C).7C). This result suggests that both Mof complexes are able to regulate Mdc1 recruitment, probably through the shared H4 K16 acetylation activity. Consistently, we found that Mdc1 was not acetylated by the Mof-Msl1v1 complex in vitro (Fig. (Fig.7D).7D). To further examine the importance of H4 K16 acetylation on Mdc1 recruitment and its binding to γH2A.X (43), we introduced FLAG-H2A.X into Mofflox/flox Cre+ cells. Mof−/− cells stably expressing FLAG-H2A.X were then derived by 4-OHT treatments to remove floxed Mof alleles. Both cell lines (Mofflox/flox and Mof−/−) were treated with 10 Gy ionizing radiation, and mononucleosomes from these cells were purified and subject to FLAG immunoprecipitation (IP). Since the majority of FLAG-H2A.X was phosphorylated after IR, FLAG IP would enrich γH2A.X-containing mononucleosomes and its binding proteins. As shown in Fig. Fig.7E,7E, H4 K16 acetylation was also enriched in γH2A.X-containing mononucleosomes after exposure to IR. Interestingly, while Mdc1 binding to dual modified mononucleosomes was clearly detected (Fig. (Fig.7E,7E, second lane), it was abolished when nucleosomes were isolated from Mof-knockout cells (Fig. (Fig.7E,7E, fourth versus second lane). These results strongly argue that H4 K16 acetylation by Mof directly affects Mdc1 binding to γH2A.X.
Mdc1 has previously been shown to bind directly to γH2A.X through its C-terminal BRCT domain (27, 43). The requirement of H4 K16ac for this interaction raised an interesting question: how is this effect achieved? It is well characterized that K16 resides in a basic region of H4 tail, which is critical for the inter- and intranucleosome interactions with an acidic pocket on the surface of H2A. The H4-H2A interaction is key for the fiber-fiber associations that contribute to formation of high-order chromatin structures in vitro (5). This interaction is regulated by H4 K16 acetylation (37, 40). Since the acidic pocket is highly conserved in H2A.X, we reasoned that if defects in Mdc1 recruitment in Mof-knockout cells are due to perturbation of interactions of H4-H2A.X, we will see a similar effect if we neutralize the charges of the acidic pocket on H2A.X. To test this hypothesis, we mutated key residues (E91 and E92) in the acidic pocket region of H2A.X to R91L92 (RL sequence). Since the RL sequence is naturally found on an H2A variant, H2A.Bbd, it does not affect the folding of mutant H2A.X. More importantly, structure studies have shown that E91 and E92 of H2A.X are not involved in the binding of the BRCT domain of Mdc1 to H2A.X (43). We stably introduced both wild-type H2A.X and the H2A.X-RL mutant into H2A.X −/− cells. Upon 10 Gy IR exposure, cells expressing exogenous H2A.X or H2A.X-RL showed indistinguishable (i.e., according to intensity and focus numbers) immunostaining for γH2A.X, suggesting that the mutant did not affect the integration of H2A.X into chromatin and its phosphorylation by ATM/ATR family kinases (Fig. (Fig.88 A). Interestingly, IRIF for Mdc1 was completely abolished in cells expressing the H2A.X-RL mutant, while it was normal in the presence of wild-type H2A.X (Fig. (Fig.8A).8A). Consistently, the level of chromatin-associated Mdc1 was also decreased in H2A.X-RL cells (Fig. (Fig.8B).8B). As controls, similar levels of H4 K16ac and γH2A.X were detected in both cell lines (Fig. (Fig.8B).8B). These results suggest that Mdc1 recruitment to DNA damage foci is likely to be controlled by trans interactions between the H4 tail and the acidic pocket of H2A.X, which are subject to regulation by Mof-mediated H4 K16 acetylation.
In this study, we have established a conditional mouse knockout model for Mof and derived 4-OHT-inducible Cre-ER MEF cell lines for Mof+/+ and Mof−/−. Using these cell lines, we demonstrate that Mof and H4 K16 acetylation are important for maintaining genome integrity. Loss of Mof and H4 K16 acetylation led to spontaneous genome instability, severe cell cycle arrest at G2/M phase (Fig. (Fig.22 and and3),3), and defects in both NHEJ- and HR-mediated DNA repair (Fig. (Fig.4).4). Importantly, although Mof deletion does not affect ATM autophosphorylation or phosphorylation of H2A.X and other targets in the ATM signaling cascade, the recruitment of repair mediator proteins Mdc1, 53bp1, and Brca1 to DNA damage foci are abolished in Mof−/− cells (Fig. (Fig.5).5). Mechanistic studies show that H4 K16 acetylation may influence γH2A.X signaling through the intra- or internucleosomal interactions with the acidic patch on H2A.X. Together, we have demonstrated a novel mechanism for the chromatin-mediated DNA damage response.
MOF is implicated in the DDR process by several earlier studies (15, 39, 48). For example, it has been reported that MOF interacts with proteins such as the MORF-related gene on chromosome 15 (MRG15) (12, 35), Ruvb1/2 (10, 20), and DNA-dependent protein kinase catalytic subunit (DNA-PKcs) (39) that are important for DNA damage repair. Cellular assays using cells treated with MOF siRNAs or with overexpression of a dominant-negative MOF mutant also show that MOF is involved in multiple steps of the DNA damage repair process (15, 39, 47). However, RNA interference-based loss of function studies for MOF in the DNA damage response sometimes give rise to conflicting results that need to be clarified. Most notably, MOF was reported to be important for IR-induced ATM and H2A.X phosphorylation at an early stage of DSB in some studies (15, 48), while it was shown to have opposite effects on these processes in others (47). These discrepancies are probably due to differences in RNA interference knockdown efficiency, the different immortalized cell lines used, and/or differences in IR conditions (i.e., dosage, time course, etc). Similar discrepancies in the Mof function in cell cycle control are also found in the literature (15, 48). Using cells carrying conditional alleles for Mof allows us, for the first time, to study the function of Mof in DNA damage repair in primary cell cultures with a well-defined genetic background. We find that Mof does not affect the earlier steps of the damage response, such as ATM signaling and γH2A.X focus formation, in response to 10-Gy IR exposure (although we cannot rule out its effects in cells treated with a low dosage of IR). Instead, it specifically affects the binding of repair mediator protein Mdc1 to γH2A.X as well as the recruitment of other downstream repair mediator proteins, such as 53bp1 and Brca1 (Fig. (Fig.6).6). Consistent with the role of Mof at a later stage of DDR, we find that an IR-induced global increase in the level of H4 K16 acetylation occurs approximately 1 h after irradiation in primary MEF cells (Fig. (Fig.5A).5A). This increase in H4 K16 acetylation is likely a result of IR-induced Mof association with chromatin but not changes in Mof expression (39). Of note, a smaller percentage of Mof-knockout cells have foci for Mdc1, 53bp1, and Brca1, which are related to spontaneous DNA damage during the normal cell cycle (data not shown). However, it remains inconclusive whether they underlie the significant spontaneous genome instability observed for these cells or are simply a result of a reduced percentage of S-phase cells in Mof−/− culture (Fig. (Fig.33).
Previous studies have shown that Mof plays important roles in transcription regulation through acetylating both H4 K16 and nonhistone substrates, such as transcription factor p53 (26). In addition, Mof also interacts with other chromatin-modifying activities that are important for establishing a euchromatic environment (10). Thus, it is important to establish whether the effects of Mof knockout on DNA damage repair are direct and whether they are due to loss of H4 K16 acetylation. Our study provides several lines of evidence to support direct roles of Mof and H4 K16 acetylation in the recruitment of key repair mediator protein Mdc1 to repair foci: (i) Mof knockout does not affect expression of the Mdc1, 53bp1, or Brca1 gene (Fig. (Fig.6C).6C). This result is consistent with a recent report that Mof knockdown in 293 cells does not affect most genes in the DNA damage repair pathway (39). (ii) The defects in Mdc1 recruitment can be rescued by expressing exogenous wild-type Mof but are rescued significantly less so by expressing the Mof mutant E350Q (Fig. 6D and E). The E350Q mutation of Mof has reduced HAT activity without affecting Mof complex formation. (iii) Upregulation of H4 K16 acetylation (Fig. (Fig.5A)5A) and its enrichment in γH2A.X-containing mononucleosomes (Fig. (Fig.7E)7E) are observed at 1 h post-IR. Importantly, dual modifications of H4 K16ac and H2A.X phosphorylation are required for Mdc1 association with isolated nucleosomes (Fig. (Fig.7E).7E). (iv) Mdc1 cannot be acetylated by the Mof-Msl1v1 complex, ruling out its regulation by Mof-mediated posttranslational modifications (Fig. (Fig.7D).7D). (v) Last, but not least, Mdc1 recruitment is regulated by both the Mof-Msl and Mof-Msl1v1 complexes (Fig. (Fig.7).7). Despite their distinct functions in transcription regulation and nonhistone substrate acetylation, both Mof complexes are important for acetylating H4 K16 in vivo (26). Taken together, our results presented here strongly argue that Mof and its H4 K16 acetyltransferase activity are directly involved in the DNA damage repair process at the step of repair protein recruitment.
The coordination between the DNA repair machinery and chromatin remodeling activities has been extensively studied (52). DNA damage elicits immediate formation of repair centers containing gigadalton-sized protein assemblies of many repair proteins (38, 46). The formation and spreading of such assemblies require a permissible chromatin environment, facilitated by ATP-dependent chromatin-remodeling activities, such as those of the INO80 complex and histone-modifying enzymes (8, 22). In particular, histone acetyltransferase Tip60 has been reported to be intensely involved in DDR (42, 45). Tip60 is a highly conserved MYST family HAT and a close homologue of MOF. It targets histone H4 (except H4 K16) and H2A tails. It was found that the Tip60 complex functions at multiple steps of DDR. It directly (i) affects ATM autophosphorylation and activation (44); (ii) facilitates γH2A.X eviction or exchange at damage sites (19, 24); (iii) recruits ribonucleotide reductase and influences DNA repair through a chromatin-independent mechanism (34); and (iv) importantly, modulates loading of 53bp1, but not Mdc1, to the damage foci after IR (32). These functions are in contrast to those of MOF, despite their sequence homology. In this study, we find that Mof-knockout cells have normal ATM autophosphorylation and γH2A.X kinetics after IR-induced DNA damage (Fig. (Fig.5).5). However, the recruitment of repair proteins Mdc1 and its downstream effectors is completely abolished (Fig. (Fig.6).6). This effect depends on the activity of Mof on H4 K16 acetylation (see above). Inactivation of the acetyltransferase activity of Mof by either mutating its HAT domain or knocking down the key regulatory components of the Mof complexes dramatically reduces focus formation by Mdc1 (Fig. (Fig.66 and and7).7). Given the different substrate specificities for Mof and Tip60 (13, 19, 51), the distinct phenotypes that we have observed in Mof-knockout cells probably reflect the unique role of H4 K16 acetylation in DNA damage repair compared to the roles of other histone acetylation events.
In contrast to other chromatin-remodeling activities involved in DNA damage repair (31), Mof is likely to influence the repair process through regulating higher-order chromatin structures. This attributes to H4 K16 acetylation the crucial role of regulating internucleosome interactions between the basic patch (amino acids 14 to 19) on the histone H4 tail and the acidic pocket on the H2A-H2B surface in chromatin fibers (5, 9, 28, 30). Biochemical and biophysical studies have shown that unacetylated H4 K16 is required to achieve the maximum propensity of nucleosome arrays to fold into secondary or tertiary chromatin structures in vitro (40). On the contrary, 30% H4 K16 acetylation alleviates compaction of 30 nm of chromatin fiber to a greater degree than deletion of the H4 N-terminal tail (37). Thus, it is highly likely that loss of Mof reduces accessibility of nucleosomes to repair proteins such as Mdc1 by increasing the compaction of chromatin structures at the damage foci. In support of this, we find that a charge-neutralizing mutation, H2A.X-RL, which abolishes the H4 tail interaction and which is shown to facilitate heterochromatination throughout the nuclei (55), recapitulates the effect of Mof knockout in Mdc1 recruitment (Fig. (Fig.8A).8A). This result suggests that Mof-mediated H4 K16 acetylation probably regulates DNA damage repair through affecting the inter- and intranucleosome interactions of H4-H2A tails, which are essential for establishing the proper higher-order chromatin structure. Our study described here also points to an interesting and unexpected role for the acidic pocket of H2A.X in the signaling cascade of DNA damage responses. It has previously been shown that the acidic pocket of H2A is used by proteins or protein complexes to associate with nucleosome particles (1, 2). Since blocking the acidic pocket by interacting proteins (e.g., LANA) has the same effect as the charge-neutralizing mutations (i.e., H2A.X-RL) (6, 55), it will be interesting to test whether proteins that bind to this region of H2A.X block the recruitment of DNA repair proteins and regulate the DNA damage response in the same way as Mof. Future studies on the acidic pocket of H2A.X and its interacting proteins are necessary to access the generality of this novel mechanism for regulating the DNA damage repair process.
We thank Thomas Saunders at the UM ES Core for helping with the generation of Mof-knockout mice and Zhenkun Lou at Mayo Clinic for providing Mdc1 antibody. We thank Melanie Adams-Cioaba and Yahong Guo in the J.M. laboratory for purifying 53bp1, Yiping Wu and Jeffrey Buis in the D.F. laboratory, and Jiaxue Wu in the X.Y. laboratory for advice on DNA repair assays.
This work was supported by NIH funds to D.F., X.Y., and Y.D. and an American Cancer Society Research Scholar grant to Y.D.
Published ahead of print on 13 September 2010.