|Home | About | Journals | Submit | Contact Us | Français|
In Saccharomyces cerevisiae, histone H3 lysine 56 acetylation (H3K56ac) occurs in newly synthesized histones that are deposited throughout the genome during DNA replication. Defects in H3K56ac sensitize cells to genotoxic agents, suggesting that this modification plays an important role in the DNA damage response. However, the links between histone acetylation, the nascent chromatin structure, and the DNA damage response are poorly understood. Here we report that cells devoid of H3K56ac are sensitive to DNA damage sustained during transient exposure to methyl methanesulfonate (MMS) or camptothecin but are only mildly affected by hydroxyurea. We demonstrate that, after exposure to MMS, H3K56ac-deficient cells cannot complete DNA replication and eventually segregate chromosomes with intranuclear foci containing the recombination protein Rad52. In addition, we provide evidence that these phenotypes are not due to defects in base excision repair, defects in DNA damage tolerance, or a lack of Rad51 loading at sites of DNA damage. Our results argue that the acute sensitivity of H3K56ac-deficient cells to MMS and camptothecin stems from a failure to complete the repair of specific types of DNA lesions by recombination and/or from defects in the completion of DNA replication.
The building block of chromatin is the nucleosome core particle, within which DNA is wrapped around a protein octamer that consists of two molecules each of histones H2A, H2B, H3, and H4. De novo nucleosome assembly occurs through DNA replication-coupled or uncoupled pathways (25, 70). During DNA replication, newly synthesized histones form complexes with chaperones such as Asf1, Rtt106, and chromatin assembly factor 1 (CAF-1), which collectively act to rapidly assemble new nucleosomes behind replication forks (23, 37, 41, 75, 100). New H3 and H4 molecules that are assembled into chromatin during replication are acetylated on multiple lysine residues (11, 34, 53, 72, 77, 78, 96). Since most of these acetylation marks are removed from chromatin after histone deposition, it is thought that these modifications play a role in chromatin assembly (33, 72). In Saccharomyces cerevisiae, histone H3 lysine 56 acetylation (H3K56ac) is present in virtually all the new H3 molecules deposited behind replication forks (12, 53). H3K56ac is mediated by the Rtt109 acetyltransferase in concert with the histone chaperone Asf1 (19, 27, 28, 67, 86), while its genome-wide removal is catalyzed by the sirtuins Hst3 and Hst4 (12, 50, 83). Originally discovered in S. cerevisiae, H3K56ac exists in other fungi, such as Candida albicans and Schizosaccharomyces pombe (26, 46, 90, 92), and is also present in human cells, but at low stoichiometry (17, 56, 84, 88, 94, 97). This is in stark contrast to S. cerevisiae, where 20 to 30% of total H3 molecules are acetylated on lysine 56 in asynchronous cells (20, 53, 91).
In S. cerevisiae, defects in replication-coupled nucleosome assembly or acetylation of new histones result in sensitivity to genotoxic agents (36, 41). Cells that lack H3K56ac (H3 K56R, asf1Δ, or rtt109Δ cells) are extremely sensitive to genotoxic agents that damage DNA during replication, but the origin of this phenotype is unclear (13, 19, 27, 31, 53, 61, 67). H3K56ac facilitates replication-coupled chromatin assembly by increasing the association of new histone molecules with CAF-1 and Rtt106 (41). This led to the suggestion that the genotoxic-agent sensitivity of cells lacking H3K56ac might result from their defect in chromatin assembly (41). In addition, H3K56ac-dependent chromatin reassembly after repair of a site-specific DNA double-strand break (DSB) has been proposed as a prerequisite for deactivation of the DNA damage checkpoint and cell survival (13, 14, 35). However, several lines of evidence suggest that the presence of K56-acetylated H3 molecules in nascent chromatin also contributes to DNA repair. First, cells lacking H3K56ac are clearly more sensitive to genotoxic agents than chromatin assembly mutants (41). Second, activation of the checkpoint kinases following exposure to genotoxic agents triggers degradation of the H3K56 deacetylase, which leads to the persistence of H3K56ac in chromatin after DNA replication (53, 83). Since overexpression of Hst3 causes genotoxic-agent sensitivity (50), it seems plausible that the rationale behind DNA damage-induced degradation of Hst3 is to promote the retention of H3K56ac in H3 molecules that are packaged into chromatin. Third, the presence of H3K56ac reduces nucleosome stability and enhances the rate of DNA end dissociation from nucleosomes, which may facilitate the action of specific enzymes at replication forks where DNA is damaged (4, 58, 76). Consistent with this, the H3 K56 acetyltransferase Rtt109 has been genetically linked to a ubiquitin (Ub) ligase complex containing the Rtt101, Mms1, and Mms22 proteins (15, 21, 69). Mutations that abolish the expression of these proteins lead to genotoxic-agent sensitivity, suggesting that H3K56ac may contribute to the regulation of their activity when DNA damage occurs at replication forks. However, the repair processes that depend on H3K56ac remain poorly understood.
In this article, we investigate how H3K56ac promotes cell survival after exposure to genotoxic agents that cause DNA damage during replication, such as methyl methanesulfonate (MMS) and camptothecin (CPT). Our results show that in the absence of H3K56ac, genotoxic-agent exposure markedly delays the completion of DNA replication and leads to persistent lesions that contain the homologous recombination (HR) protein Rad52. Taken together, our data suggest that the presence of H3K56ac in nascent chromatin is important for cells to complete the repair of DNA lesions that occur during DNA replication.
Yeast cells were grown using standard techniques in yeast extract-peptone-dextrose (YPD) or synthetic medium. Microscopy experiments were conducted using yeast grown in synthetic medium supplemented with 100 μg/ml adenine to reduce autofluorescence. Unless indicated otherwise, the experiments were performed at 30°C. For spot assays, 5-fold serial dilutions of saturated overnight cultures were spotted onto the medium. Experiments involving G1 synchronization were carried out by incubating exponentially growing cells in a medium containing 5 μg/ml α-factor for 1.5 h at 30°C, followed by the addition of a second dose of α-factor (5 μg/ml) for 1 h at 30°C. bar1Δ cells were arrested in a similar way except that 50 ng/ml α-factor was used. Cells were then released from the G1 arrest in a medium containing 50 μg/ml of pronase.
Pulsed-field gel electrophoresis (PFGE) was performed as described previously (52). A total of 2 × 107 cells per sample were used to make the agarose plugs, and half of the plugs were loaded into the gel. Densitometry was performed using MultiGauge software, version 3.0 (Fujifilm).
Wild-type (WT) and rtt109Δ cells were engineered to incorporate halogenated nucleosides by using integration cassettes encoding a human equilibrative nucleoside transporter (hENT1) and a herpes simplex virus thymidine kinase gene as described previously (89). Cells were synchronized in G1 by the addition of 5 μg/ml α-factor for 90 min at 30°C, followed by the addition of a second dose of α-factor (5 μg/ml) for 75 min at 30°C. In the last 15 min of the synchronization, 40 μg/ml bromodeoxyuridine (BrdU) was added to the culture. The cells were then released into S phase in the presence of 400 μg/ml BrdU and 0.03% MMS for 1.5 h. Cells were washed in YPD–2.5% sodium thiosulfate to inactivate the MMS and were then resuspended in fresh medium containing 400 μg/ml BrdU. Genomic DNA was extracted in low-melting-point agarose plugs and was stained with YOYO-1 (Molecular Probes). DNA was resuspended in 50 mM 2-(N-morpholino)ethanesulfonic acid (MES), pH 5.7, to a final concentration of 150 ng/ml. DNA fibers were stretched on silanized coverslips as described previously (47) and were denatured for 25 min in 1 M NaOH. BrdU was detected with a rat monoclonal antibody (clone BU1/75; AbCys) and a secondary anti-rat IgG coupled to Alexa 488 (Molecular Probes). DNA molecules were counterstained with an antibody against guanosine (Argene) and an anti-mouse IgG coupled to Alexa 546 (Molecular Probes). Images were recorded with a Leica DM6000 B microscope coupled to a CoolSNAP HQ charge-coupled device camera (Roper Scientific) and were processed as described previously (62). BrdU signals and DNA fibers were measured using MetaMorph, version 6.2 (Universal Imaging Corp.).
Immunoblotting was performed using standard molecular biology techniques. Yeast whole-cell lysates were prepared as described previously (38). For Rad53-Flag immunoblotting and Rad53 autophosphorylation assays, a standard trichloroacetic acid and glass-bead lysis protein extraction protocol was used and was found to be necessary to obtain adequate yields of high-molecular-weight proteins. A custom-made polyclonal rabbit antibody (AV135) raised against H2A serine 128 phosphorylation (referred to below as H2A S128-P) was used. A mouse monoclonal antibody against the Flag epitope was purchased from Sigma (anti-Flag M2 monoclonal antibody).
In situ Rad53 autophosphorylation assays were performed as described previously (63).
The rates of spontaneous ADE2 marker loss from ribosomal DNA (rDNA) were measured as described previously (55).
Asynchronous cells were processed for fluorescence microscopy as described previously (45). Fluorophores were visualized using band-pass cyan fluorescent protein (CFP; catalog no. 31044 v2), yellow fluorescent protein (YFP; catalog no. 41028), and red fluorescent protein (RFP; catalog no. 41002c) filter sets from Chroma (Brattleboro, VT) (see Fig. 7). For experiments involving transient exposure to a genotoxic agent, cells were grown in synthetic medium supplemented with adenine throughout the experiments to minimize autofluorescence. Cells were fixed for 30 min in 0.1 M potassium phosphate buffer (pH 6.4) containing 3.7% formaldehyde at the time points of interest and were then washed in 0.1 M potassium phosphate buffer. Fixed cells were kept at 4°C pending microscopy analysis. Cells were imaged on polylysine-coated slides with 4′,6-diamidino-2-phenylindole (DAPI) mounting medium. For time lapse imaging, cells were placed on pads containing 25% gelatin in synthetic medium plus 100 μg/ml adenine and were imaged for 9 to 12 h in a controlled-environment incubator at 25°C. Images were taken every 15 or 20 min depending on the experiment. Microscopy was performed on a DeltaVision instrument (Applied Precision, Issaquah, WA) using a 60× objective (numerical aperture [NA], 1.42). Fluorophores were visualized using band-pass YFP and DAPI filters from the standard filter set of Applied Precision. Microscopy images were analyzed using the SoftWoRx software suite, version 4.0.0, from Applied Precision, Adobe Photoshop CS4, and ImageJ 1.44P.
Four individual colonies of each genetic background were inoculated into synthetic medium lacking arginine and containing 0.0005% MMS and were grown at 30°C for 18 h. Cells were then diluted and plated on either YPD (to evaluate plating efficiency) or synthetic medium lacking arginine but containing 60 μg/ml canavanine. The experiment was performed twice, and the data presented in Fig. 6D represent the means of the pooled results of these two experiments.
DNA replication and recombination intermediates were studied as described previously (42). DNA was digested using the EcoRV and HindIII restriction enzymes. Southern blotting was performed using a probe specific to the ARS305 replication origin. The probe sequence is available on request.
Cells lacking H3K56 acetylation are sensitive to a number of genotoxic agents that cause DNA damage during replication but not to an acute exposure to ionizing radiation (19, 53). MMS and CPT cause DNA lesions during replication (60, 65, 82, 93), whereas hydroxyurea (HU) slows down replication fork progression (3, 95). We characterized the responses of cells lacking H3K56ac to these genotoxic agents. Chronic exposure to MMS or CPT strongly impaired the formation of colonies by H3 K56R or rtt109Δ mutants (Fig. 1A). In contrast, a high concentration of HU (100 mM) did not result in a severe loss of viability for mutants lacking H3K56ac; prolonged incubation of the plates led to the emergence of slow-growing colonies (Fig. 1A). To confirm these results, equal numbers of cells were plated onto YPD medium containing MMS, CPT, or HU, and CFU were counted. In agreement with the results of our spot assays, H3 K56R and rtt109Δ cells were extremely sensitive to MMS and CPT but gave rise to numbers of colonies similar to those of WT cells on plates containing HU (Fig. 1B). In contrast, mec1Δ sml1Δ cells were incapable of forming colonies on HU-containing plates (Fig. 1A and B), a finding consistent with their known sensitivity to replication fork pauses (73). We next determined the survival of asynchronous WT, rtt109Δ, and H3 K56R cells after transient exposure to MMS or HU. A low concentration of MMS led to a time-dependent decrease in the viability of rtt109Δ and H3 K56R cells, whereas incubation in 200 mM HU for as long as 6 h did not result in pronounced cell death (Fig. 1C). We also tested whether transient exposure to HU specifically during S phase was cytotoxic to mutants lacking H3K56ac. For this purpose, cells were synchronized in G1, released into S phase, and incubated in the presence of increasing concentrations of HU for 1.5 h, and their viability was assessed using colony formation assays. H3 K56R and rtt109Δ mutants were as viable as wild-type cells after a transient exposure to a high dose of HU (200 mM) (Fig. 1D).
The genetic interaction profile of RTT109 is similar to those of RTT101, MMS1, and MMS22 (15). These genes encode subunits of a Ub ligase that promotes the completion of DNA replication after MMS exposure (21, 98). Intriguingly, RTT109 deletion causes defects in the Rtt101-dependent recruitment of Rtt107 (another protein necessary for the completion of DNA replication after MMS exposure) to chromatin in response to MMS, but H3 K56R or ASF1 null mutations do not (69, 71). This has led to speculation that, in addition to H3K56, Rtt109 may acetylate other targets to regulate the Rtt101-Mms1-Mms22 complex. Our previous mass spectrometry results clearly demonstrated that H3K56 is the main target of Rtt109 in core histones (20, 81), but we could not exclude the possibility that Rtt109 might also acetylate other proteins to promote the survival of cells treated with MMS. To study this in more detail, strains carrying mutations of H3K56, RTT101, MMS1, and MMS22, either alone or in combination with an rtt109Δ mutation, were transiently exposed to MMS during S phase, and cell survival was assayed. Our results indicate that deletion of RTT109 does not enhance the MMS sensitivity of H3 K56R mutant cells (Fig. 1E, top left). Similar results were obtained in spot assays (data not shown). This epistasis experiment strongly suggests that H3K56 is an important substrate of Rtt109 in response to MMS. Our results also show that deletion of RTT101, MMS1, or MMS22 does not exacerbate the MMS sensitivity of rtt109Δ mutant cells (Fig. 1E), implying that the Rtt101-Mms1-Mms22 Ub ligase, Rtt109, and H3K56ac function in the same genetic pathway to confer resistance to MMS.
MMS-induced lesions block the progression of DNA replication forks because replicative polymerases cannot accommodate 3-methyladenine in their catalytic sites (10, 24, 39). This can result in stalled or collapsed forks and prevent the completion of DNA replication. We performed DNA combing to determine whether cells lacking H3K56ac could complete DNA replication after transient exposure to MMS (Fig. 2A and B) (see Materials and Methods) (89). We found that the fraction of total DNA that was replicated in WT cells increased progressively from 30 to 70% after removal of MMS from the medium (Fig. 2B, left). In contrast, during 4.5 h following MMS removal, the fraction of DNA that was replicated in rtt109Δ cells increased only slightly, from 40 to 50% (Fig. 2B). Consistent with these findings, the sizes of unreplicated DNA gaps devoid of BrdU staining were relatively constant from 90 to 270 min after MMS removal from rtt109Δ cells, but over the same period, unreplicated DNA gaps progressively decreased in size in WT cells (Fig. 2B, right).
To support our DNA combing results, we compared the recovery of pulsed-field gel electrophoresis (PFGE) signals when asynchronous populations of WT and rtt109Δ cells were transiently exposed to MMS and HU (Fig. 2C). Chromosomes containing DNA replication intermediates cannot enter PFGE gels (71). Consistent with the fact that rtt109Δ cells do not lose viability after a transient exposure to HU, we observed a recovery of PFGE signals in both WT and rtt109Δ cells between 90 and 180 min after removal of HU from the medium (Fig. 2C). We also monitored histone H2A serine 128 phosphorylation (H2A S128-P), a modification that is catalyzed by the checkpoint kinases Mec1 and Tel1 in response to DNA damage (18). Exposure to HU did not lead to an increase in the level of H2A S128-P in either WT or rtt109Δ cells (Fig. 2C). In contrast to the HU results, exposure of rtt109Δ cells to MMS led to a persistent loss of the PFGE signal, suggesting that DNA replication could not be completed in H3K56ac-deficient cells (Fig. 2C). MMS-induced H2A S128-P also persisted for at least 180 min after MMS removal from rtt109Δ cells (Fig. 2C).
We also assayed the effect of transient exposure to MMS during S phase by PFGE. Even 360 min after MMS removal, rtt109Δ cells displayed only partial recovery of the PFGE signal (Fig. 2D and E). In contrast, the intensity of the PFGE signal increased beyond G1 levels in WT cells (Fig. 2D and E), presumably reflecting the duplication of DNA content. Compared to that for other chromosomes, we found that completion of chromosome XII replication was particularly compromised in rtt109Δ cells (Fig. 2F). This chromosome contains the RDN1 locus, which consists of 100 to 200 direct repeats (about 1 to 2 Mb in total) of ribosomal DNA (rDNA). This suggests that the rDNA locus sustain persistent damage in MMS-treated cells lacking H3K56ac. Taken together, our results strongly suggest that transient exposure to MMS irremediably compromises the completion of DNA replication in cells lacking H3K56ac.
Cells lacking H3K56ac have been reported to be defective in DNA damage checkpoint deactivation after the repair of a DSB by single-strand annealing, leading to a permanent arrest in G2 even after the completion of DNA repair (13, 14). Our colony formation assays showed that cells lacking H3K56ac did not lose viability after a transient exposure to HU (Fig. 1B), which is known to strongly activate the intra-S-phase DNA damage checkpoint. This argues that cells devoid of H3K56ac can recover from HU-induced checkpoint activation. To explore this further, we studied the activation and deactivation of the DNA damage checkpoint in WT and rtt109Δ cells. Cells synchronized in G1 were released into the cell cycle in the presence of MMS. Under these conditions, H2A S128-P mediated by Mec1/Tel1 and the kinase activity of Rad53 were induced with comparable kinetics in WT and rtt109Δ cells (Fig. 3A). Deactivation of the DNA damage checkpoint was monitored after a transient exposure to HU or MMS during S phase. We detected strong Rad53 autophosphorylation immediately after the removal of genotoxic agents (Fig. 3B and C, 0 min). Interestingly, after the removal of HU, the kinase activity of Rad53 decreased in both WT and rtt109Δ cells, albeit with slower kinetics in rtt109Δ cells (Fig. 3B). However, we observed that Rad53 activity was elevated even in rtt109Δ cells that were not treated with genotoxic agents (Fig. 3B and C, α-factor). This may explain why Rad53 activity never completely vanished as a function of time after HU removal (Fig. 3B). We also noticed a moderate upward shift in the electrophoretic mobility of Rad53-Flag up to at least 360 min after the removal of HU from rtt109Δ cultures, even though our in situ Rad53 autophosphorylation assays revealed that the kinase activity of Rad53 was reduced to basal levels much more rapidly (Fig. 3B). This may be explained by the fact that phosphorylation of Rad53 occurs at multiple sites in the protein, but only the phosphorylation of key residues is needed for kinase activity (79). In contrast to the HU results, transient exposure to MMS induced persistent activity of Rad53 in rtt109Δ cells, as evidenced by in situ autophosphorylation assays and the slower electrophoretic mobility of Rad53-Flag (Fig. 3C). Taken together, these results demonstrate that following HU removal, cells lacking H3K56ac are capable of deactivating the DNA damage checkpoint kinase Rad53.
To further investigate the role of H3 K56ac in regulating checkpoint activity in response to genotoxic agents, we tested the effect of pharmacological inhibition of DNA damage checkpoint kinases in rtt109Δ cells transiently exposed to MMS. Caffeine is an inhibitor of Mec1 and Tel1, which are key kinases in the DNA damage checkpoint. If a failure to deactivate the checkpoint after MMS exposure were the main cause of lethality in cells lacking H3K56ac, one would expect that pharmacological inhibition of the checkpoint kinases might attenuate their MMS sensitivity. To test this, asynchronous cells were incubated in MMS for 2 h. MMS was then inactivated, and cells were incubated for 1 h in YPD to allow recovery from DNA damage under conditions where checkpoint signaling is functional (Fig. 3D). Cells were then either treated with 0.5% caffeine for 30 min or left untreated (Fig. 3D). Immunoblotting of Rad53-Flag clearly showed that caffeine treatment significantly reduced the levels of the slower-migrating phosphorylated forms of Rad53 in both WT and rtt109Δ cells (Fig. 3D, immunoblot, lanes 30 min CAF and 30 min CTRL), suggesting that Rad53 activity had been attenuated. Importantly, the viability of WT cells was not reduced by caffeine under our specific experimental conditions (Fig. 3D, bar graph). Thus, as long as checkpoint kinases were active during MMS treatment, subsequent inhibition of Mec1 and Tel1 by caffeine did not irreversibly damage DNA replication forks. Under the same conditions, the survival of rtt109Δ cells was not improved by caffeine treatment (Fig. 3D, bar graph). These results argue that defects in DNA damage checkpoint deactivation are unlikely to account for the loss of viability of H3K56ac-deficient cells exposed to MMS.
Homology-dependent processes are involved in the repair of DNA damage at replication forks through multiple pathways (10). Rad52 is an important HR protein that forms intranuclear foci upon DNA damage (2, 44, 45). The persistence of these foci as a function of time may indicate a problem in completing homology-dependent recombination. We investigated the dynamics of Rad52-YFP foci in WT and H3 K56R cells following transient exposure to genotoxic agents (Fig. 4 and and5).5). After MMS exposure during S phase, H3 K56R cells suffered a 90% loss of viability, compared with only about a 9% loss for WT cells (Fig. 4B). As reported previously, an α-factor-arrested population of cells lacking H3K56ac exhibited an abnormally high frequency of spontaneous Rad52-YFP foci compared with that in WT cells (9% versus 0.5%, respectively [Fig. 4C]) (27). Immediately after the removal of MMS, when the checkpoint was active (Fig. 3C), the fraction of cells with Rad52-YFP foci was low, essentially the same as that in G1 (Fig. 4C, 0 min). This is consistent with published results showing that activation of the intra-S-phase DNA damage checkpoint precludes the formation of Rad52 foci (1, 5). However, 60 min after the quenching of MMS, when cells had reached a 2C DNA content (Fig. 4A), 60% of WT and 78% of H3 K56R cells displayed Rad52-YFP foci (Fig. 4C). At 120 min after MMS removal, the fraction of WT cells containing Rad52-YFP foci began to decline, whereas foci continued to accumulate in H3 K56R cells until they were present in 94% of the cells (Fig. 4C). For the remainder of the time course, the fraction of WT cells with Rad52-YFP foci decreased steadily over time, while the fraction of H3 K56R mutant cells with these foci remained between 94 and 99%.
To investigate the persistence of Rad52-YFP foci, we performed time lapse microscopy on WT and H3 K56R cells that were first synchronized in G1 and then released into S phase in the presence of MMS (Fig. 4D). After 90 min, MMS was inactivated, and cells placed on gelatin pads were imaged at 25°C. We found that about 60% of WT cells formed Rad52-YFP foci that vanished within 5 h after the removal of MMS. In striking contrast, 96% of H3 K56R cells containing Rad52-YFP foci still harbored foci 10 h after MMS removal (Fig. 4D). These results suggest that a transient exposure to MMS results in persistent structures that contain Rad52 and cannot be resolved in H3 K56R cells. We determined the localization of persistent Rad52-YFP foci relative to nuclear DNA stained with DAPI in large budded G2/M cells (Fig. 4E). We observed three categories of cells (Fig. 4E, left). In the first category, only the mother cell had a well-defined area of DAPI staining with superimposed Rad52-YFP foci. The second category consisted of cells in which DAPI staining was present mostly at the bud neck, with superimposed Rad52-YFP foci. The third category included cells in which there were clearly separated DAPI-stained areas in each daughter cell and Rad52-YFP foci within either one or both of the DAPI areas. We quantified the fraction of cells with Rad52-YFP foci that belonged to the second and third categories as a function of time after MMS removal. A larger fraction of H3 K56R cells than of WT cells showed DAPI staining at the bud neck with superimposed Rad52-YFP foci. In H3 K56R cells, this fraction gradually increased as a function of time to reach more than 60% at 240 min following MMS removal (Fig. 4E). Interestingly, at the next time point (360 min after MMS removal), the fraction of H3 K56R cells displaying Rad52-YFP foci and DAPI staining at the bud neck had decreased to roughly 40%, whereas the fraction of cells with Rad52-YFP foci superimposed on either of two clearly separated masses of DAPI-stained DNA had increased (Fig. 4E). This result suggests that despite the presence of persistent structures containing Rad52, cells slowly proceed to segregate DNA into daughter cells.
Rad52 acts early during HR, and Rad52 foci can contain other recombination proteins, depending on which step of HR is occurring at sites of DNA damage (44). An important function of Rad52 is to displace replication protein A (RPA) from single-stranded DNA to permit the formation of Rad51 nucleoprotein filaments. The Rad51 filaments then promote homology searching and strand invasion of the homologous template DNA (80). We determined the dynamics of Rad51 foci after a transient exposure of cells lacking H3K56ac to MMS during S phase (Fig. 4F). We found that the YFP-RAD51 allele was hypomorphic and conferred MMS sensitivity. Because of this, we used lower concentrations of MMS than those in previous experiments (0.01%). As a control, we verified that following transient exposure to 0.01% MMS, the fraction of rtt109Δ cells displaying Rad52-YFP foci remained high for at least 8 h after the removal of MMS (data not shown). Thus, rtt109Δ and H3 K56R cells behaved similarly with regard to the persistence of Rad52-YFP foci following MMS treatment. Cells were synchronized in G1 and were released into the cell cycle in the presence of 0.01% MMS for 1 h 40 min. MMS was inactivated, and live cells were imaged on gelatin pads to determine the duration of YFP-Rad51 foci. We found that approximately 60% of WT cells displayed long-lasting YFP-Rad51 foci (>9.5 h), even though these cells lost only about 10% viability after exposure to 0.01% MMS (Fig. 4F). This may be due to the fact that the YFP-RAD51 allele is hypomorphic. In contrast to those in WT cells, essentially all (98%) YFP-Rad51 foci present in rtt109Δ cells lasted more than 9.5 h. These results indicate that cells lacking H3K56ac are capable of forming Rad51 foci after MMS exposure, suggesting that an HR step downstream of Rad51 is impaired in rtt109Δ cells.
In addition to MMS, cells lacking H3K56ac are also sensitive to CPT (53). The initial DNA lesion caused by CPT is rather different from those generated by MMS. Inhibition of DNA topoisomerase I (Top1) by CPT converts single-strand nicks covalently linked to Top1 into DNA double-strand breaks during replication. We investigated the formation of Rad52-YFP foci that arose from transient exposure to CPT (Fig. 5A to D). Cells were arrested in G1 and were released into the cell cycle in a medium containing 10 μg/ml CPT for 1.5 h. After the removal of CPT, H3 K56R cells accumulated with a 2C DNA content, whereas WT cells resumed cycling between 120 and 240 min (Fig. 5A). Transient exposure to CPT during S phase led to an important decrease in the viability of H3 K56R cells under these conditions (Fig. 5B). In contrast to MMS (Fig. 4C, 0 min), CPT resulted in the formation of Rad52-YFP foci immediately during treatment in both WT and H3K56R cells (Fig. 5C, 0 min). This may be due to the fact that CPT does not strongly induce the DNA damage checkpoint during S phase (1, 68). As was the case with MMS, the fractions of WT and H3 K56R cells containing Rad52-YFP foci were initially similar (Fig. 5C, 0 min). However, starting at 60 min after removal of CPT, the fraction of H3 K56R cells containing Rad52-YFP foci remained constant at around 92% for as long as 360 min, whereas in WT cells this fraction declined over time to reach about 28% (Fig. 5C). We evaluated the duration of individual Rad52-YFP foci after a transient exposure to CPT by time lapse microscopy (Fig. 5D). In WT cells, most Rad52-YFP foci were short-lived, whereas 80% of the foci formed in H3 K56R cells were still present after 10 h. This result demonstrates that, as is the case for MMS, DNA damage generated by transient exposure to CPT during S phase gives rise to persistent structures that contain Rad52.
Our earlier experiments showed that, unlike DNA damage caused by MMS or CPT, replication fork pausing generated by transient exposure to HU did not lead to a major loss of viability in cells lacking H3K56ac (Fig. 1C). In spite of this, the presence of H2A S128-P after transient exposure to HU during S phase in rtt109Δ cells suggested that HU causes DNA damage in a fraction of cells (Fig. 3B). One possibility to reconcile these observations would be that the fraction of cells incurring DNA damage or the number of lesions per cell is lower after treatment with HU than after treatment with MMS or CPT, thus resulting in severe lethality with MMS or CPT, but not with HU. To address these issues, we monitored Rad52-YFP foci in WT and H3 K56R cells transiently exposed to 200 mM HU during S phase (Fig. 5E to H). Cells synchronized in G1 were released into S phase in the presence of 200 mM HU for 1.5 h and were then incubated in fresh medium lacking HU. Based on DNA content, most WT and H3 K56R cells resumed cycling after a transient exposure to HU (Fig. 5E). As was the case for rtt109Δ cells (Fig. 1C), transient exposure to HU did not result in a loss of viability of H3 K56R cells (Fig. 5G). Rad52-YFP foci were absent immediately after HU removal (Fig. 5F, 0 min), presumably because HU strongly activates the intra-S-phase DNA damage checkpoint (Fig. 3B). However, 60 min after HU removal, we observed that about 20% of WT cells and 37% of H3 K56R cells contained Rad52-YFP foci (Fig. 5F). This fraction is markedly lower than that with CPT or MMS treatment. Interestingly, at the next time point (120 min after HU removal), WT cells were essentially devoid of Rad52-YFP foci, and this remained the case until at least 360 min (Fig. 5F). In contrast, the proportion of H3 K56R cells containing Rad52-YFP foci slowly declined over time, but 22% of cells still contained foci at 360 min after HU removal (Fig. 5F). We measured the duration of Rad52-YFP foci in living cells transiently exposed to HU and found that in WT cells, most Rad52-YFP foci (89%) vanished within an hour of their appearance (Fig. 5H). While 40% of HU-induced Rad52-YFP foci formed in H3 K56R cells also disappeared rapidly, 36% of them lasted for at least 8 h (Fig. 5H). We conclude that transient exposure to HU during S phase also results in some persistent Rad52 foci in H3 K56R cells but that a major fraction of HU-induced foci are short-lived. Since most mutant cells survived a transient exposure to HU (Fig. 5G), it seems likely that, given sufficient time, most HU-induced foci would eventually be resolved in H3 K56R cells. This would be consistent with the slow proliferation phenotype of H3 K56R cells observed during chronic exposure to HU (Fig. 1A). These results suggest that, regardless of the nature of the genotoxic agent, DNA damage during replication leads to the formation of persistent Rad52 foci in at least a fraction of H3 K56R cells.
The MMS sensitivity of cells lacking H3K56ac could be due to defects in the base excision repair (BER) pathway, which is important for the removal of MMS-damaged DNA bases, such as 3-methyladenine (54). In addition, BER intermediates could, in principle, be converted to DNA DSBs during replication. To assess the relationship between the BER pathway and H3K56ac, we deleted BER genes, either alone or in combination with RTT109, and evaluated the MMS sensitivities of the resulting yeast strains. We found that BER mutants (mag1Δ, apn1Δ, or apn2Δ mutants) were far less sensitive to MMS than rtt109Δ cells and that BER mutations augmented the sensitivity of rtt109Δ mutants to low doses of MMS (Fig. 6A), arguing that the main function(s) of Rtt109 in response to MMS is not to promote BER. To confirm this, we investigated whether cells lacking H3K56ac were able to repair MMS-induced DNA damage in G1, when MMS lesions are repaired by BER (49). We took advantage of the fact that MMS generates heat-labile DNA lesions. Because of this, PFGE of MMS-treated yeast chromosomes results in a smear of broken DNA molecules when samples are incubated at high, but not at low, temperatures (48, 49). Cells lacking the Bar1 protease were arrested in G1 using α-factor, and MMS was added for 1 h. MMS was then inactivated, and cells were incubated in fresh medium containing α-factor but lacking MMS (Fig. 6B). For proteinase K treatment, PFGE samples were divided into two aliquots: one aliquot was incubated overnight at 65°C, while the other was incubated at 30°C. We found that heat-labile DNA lesions were removed equally well from WT and rtt109Δ cells, since the PFGE signal was recovered within 30 min after quenching of the MMS in both strains (Fig. 6B). Some DNA smearing remained in both the 30°C and 65°C samples, indicating that DNA DSBs were infrequently produced by MMS exposure, as previously reported (49). We conclude that repair of MMS-induced lesions in G1 is indistinguishable in WT and rtt109Δ cells, which strongly suggests that BER is functional in cells lacking H3K56ac.
Another important pathway that confers resistance to MMS is Rad6- and Rad18-dependent DDT (57, 99). DDT is subdivided into two main branches: error-prone translesion synthesis and error-free DNA lesion bypass (99). To address whether Rtt109 and H3K56ac are involved in these pathways, we deleted RTT109 in mutants defective in DDT and evaluated the MMS sensitivities of the resulting strains. Rad6 and Rad18 form an E2-E3 complex that monoubiquitylates the DNA polymerase processivity factor PCNA to promote translesion synthesis by error-prone DNA polymerases that can accommodate alkylated DNA bases in their catalytic sites at the cost of increased mutagenesis rates (24, 39, 99). We found that deletion of either RAD6 or RAD18 led to greater sensitivity to MMS than an rtt109Δ single mutation (Fig. 6C, 0.001% MMS). In addition, deletion of RTT109 enhanced the MMS sensitivity of rad6Δ mutants (Fig. 6C, 0.00025% MMS), suggesting that H3K56ac is important for the repair of MMS-induced DNA lesions even in the absence of DDT. To study the role of Rtt109 in MMS-induced error-prone translesion synthesis, we measured the frequency of mutagenesis at the CAN1 locus by growing cells to saturation in the presence of MMS and plating them onto a canavanine-containing medium. Only cells that inactivate CAN1 can proliferate in the presence of canavanine (16, 22). MMS-induced CAN1 mutations occurred equally frequently in WT and rtt109Δ cells (Fig. 6D). Because rtt109Δ cells were reported to have an elevated frequency of gross chromosomal rearrangements (19), it was possible that the majority of MMS-triggered CAN1 mutations in rtt109Δ mutants were due to CAN1 gene alterations, rather than being point mutations introduced by error-prone DNA polymerases, such as Rev3 (a subunit of DNA polymerase zeta). To address this possibility, we compared the frequencies of MMS-induced CAN1 mutagenesis in rev3Δ versus rev3Δ rtt109Δ mutants and found that they were very similar (Fig. 6D). If the CAN1 mutations detected in rtt109Δ cells had been generated mainly through chromosomal rearrangements rather than point mutations, we would have expected the frequency of CAN1 mutagenesis in rev3Δ rtt109Δ cells to be higher than that observed in rev3Δ single mutants. Thus, these results suggest that even in rtt109Δ cells, most CAN1 mutations caused by MMS arise through error-prone translesion synthesis.
The attachment of lysine 63-linked Ub chains to PCNA by Mms2-Ubc13 (E2) and Rad5 (E3) drives error-free DNA lesion bypass through a DNA template-switching mechanism that depends on HR proteins such as Rad51 and Rad52 (8, 9, 30). Deletion of RAD52 channels MMS lesions into the error-prone translesion synthesis pathway, because it prevents Rad52-dependent template switching (22). The fact that rtt109Δ cells did not produce more canavanine-resistant colonies than WT cells after MMS treatment therefore suggests that Rad52-dependent DNA template switching is functional in the absence of H3K56ac. In addition, deletion of RTT109 in rad52Δ cells did not decrease the elevated level of CAN1 mutagenesis observed in a rad52Δ single mutant (Fig. 6D). This should not have been the case if Rtt109 had made a substantial contribution to error-prone translesion synthesis. This result suggests that tolerance of MMS-induced DNA lesions through error-prone translesion synthesis occurs with comparable efficiencies in WT and H3K56ac-deficient cells. Cells lacking RAD5 were slightly more sensitive to MMS than rtt109Δ cells, whereas deletion of RTT109 increased the MMS sensitivity of rad5Δ mutants (Fig. 6C, 0.001% MMS). Since the rad5Δ mutation likely abolishes template switching, the enhanced sensitivity of rad5Δ rtt109Δ cells suggests that H3K56ac plays a role in the response to MMS that is independent of template switching. To verify this and to investigate DNA structures that may form at MMS-damaged DNA replication forks in cells lacking H3K56ac, we performed neutral 2D gel electrophoresis. Cells were synchronized in G1 and were released into the cell cycle in the presence of 0.033% MMS for 1 h 30 min. DNA was then extracted and processed for 2D gel electrophoresis. To compare our results with those of previous studies, we probed for DNA structures located near the ARS305 origin of replication (7, 43, 51, 87). In this 2D gel system, X-shaped structures (Fig. 6E, arrows) correspond to unresolved strand invasion recombination intermediates (51). Sgs1 is known to disrupt Rad51-dependent template-switching intermediates (87). Because of this, sgs1Δ cells accumulate X-shaped structures after MMS exposure (Fig. 6E, arrows). We found that rtt109Δ cells did not produce X-shaped DNA structures after MMS treatment (Fig. 6E). Consistent with the results of our spot tests (Fig. 6C) and CAN1 mutagenesis assays (Fig. 6D), which suggested that Rtt109 is not needed for MMS-induced DNA template switching, we found that sgs1Δ rtt109Δ double mutants were able to form X-shaped structures to an extent comparable to that of sgs1Δ cells (Fig. 6E, arrows). These results provide additional evidence that cells lacking H3K56ac are proficient in Rad51/Rad52-dependent DNA template switching triggered by MMS.
The fact that cells lacking H3K56ac accumulate persistent Rad51 and Rad52 foci, but do not form X-shaped structures, suggests that the recruitment of Rad51 and Rad52 does lead to the formation of HR intermediates following MMS-induced DNA damage. We also addressed the question of whether the contribution of Rtt109 in response to MMS and CPT could be explained solely by a role in promoting Rad51- and Rad52-dependent homology-directed repair. At concentrations of MMS and CPT that were not cytotoxic to cells lacking H3K56ac, an rtt109Δ mutation enhanced the sensitivities of rad51Δ and rad52Δ mutants to both MMS and CPT (Fig. 6F). Thus, in the absence of Rad51 and Rad52, Rtt109 still confers some protection against DNA damage generated by MMS or CPT. To determine whether DSBs that are not induced by MMS or CPT led to persistent Rad52-YFP foci in H3K56ac-deficient cells, we studied the response to a single HO-induced DNA DSB at the MAT locus (Fig. 6G to I). For these experiments, we used strains that harbor a centromeric plasmid expressing the HO endonuclease under the control of a galactose-inducible promoter (29). Galactose was added to asynchronous cells grown in synthetic medium containing raffinose, and Rad52-YFP foci were examined at 60 and 240 min following galactose addition. We found that H3 K56R cells did not lose viability after HO-induced DSB formation at the MAT locus (Fig. 6G and H). Consistent with this, relative to the fraction of cells exhibiting spontaneous foci prior to galactose addition (0 min), we did not observe a significant increase in Rad52-YFP foci after HO induction in H3 K56R cells (Fig. 6I). This result confirms a previous study showing that cells devoid of H3K56ac are able to complete mating type switching with kinetics comparable to those of WT cells (13), and it demonstrates that cells lacking H3K56ac are not generally defective in all forms of HR-mediated DSB repair.
Asynchronous populations lacking H3K56ac display an anomalously high fraction of cells with spontaneous Rad52-YFP foci (2, 27) (Fig. 6I, 0 min). We compared spontaneous foci of key DNA damage checkpoint and recombination proteins in asynchronous WT and H3 K56R cells. In WT cells, spontaneous Mre11, Ddc2, and Rad52 foci were observed in 4%, 12%, and 6% of exponentially growing cells (Fig. 7A and B). In contrast, Ddc2 and Rad52 foci were both present in 29% of H3 K56R cells, whereas the frequency of Mre11 foci was essentially the same as that in WT cells (Fig. 7B). The increased frequency of Rad52 foci indicates that a greater percentage of asynchronous H3 K56R cells than of WT cells are engaged in homology-dependent processes. However, the equal numbers of Mre11 foci in WT and H3 K56R mutant cells suggest that the frequency of DSB formation does not increase in the mutant. This is because Mre11 foci are transient, and therefore, their presence reflects the rate of DSB formation rather than the total number of persistent lesions present at any given time. The observation of spontaneous foci suggests that H3 K56R mutant cells have a defect in the repair of spontaneous DNA damage, leading to the persistence and accumulation of structures that are recognized by Rad52 and Ddc2.
Because rtt109Δ cells display problems in the completion of chromosome XII duplication after MMS exposure (Fig. 2F), we reasoned that the rDNA locus might also be susceptible to the formation of spontaneous Rad52-YFP foci. To address whether H3 K56R cells accumulate Rad52 foci at the rDNA locus, the localization of Rad52-YFP foci was determined relative to that of the nucleolus marker Nop1-CFP. In WT cells, the majority of spontaneous Mre11, Ddc2, and Rad52 foci were found outside the nucleolus (Fig. 7B). In contrast, the proportion of Rad52 foci near or within the nucleolus was higher in H3 K56R mutants than in WT cells (Fig. 7B). However, the higher proportion of perinucleolar or nucleolar foci was not sufficient to account for the overall increase in Rad52-YFP foci observed in H3 K56R cells. Therefore, either DNA lesions accumulated genome-wide or rDNA lesions relocalized outside the nucleolus during repair, as described previously (85). To address whether rDNA lesions accounted for a substantial fraction of the spontaneous damage observed in H3 K56R mutants, the localization of spontaneous Rad52-YFP foci was examined in a sumoylation-defective rad52-K43R K44R K253R mutant; this form of Rad52, unlike wild-type Rad52, is not excluded from the nucleolus when DNA lesions occur within RDN1 (85). In H3 K56R cells, the rad52-K43R K44R K253R mutant caused an increase in the overall fraction of cells with Rad52-YFP foci (from 29% up to 38% [Fig. 7B]) and a significant shift in the localization of those foci toward the nucleolus (P, 9.9 × 10−8 by a one-tailed Fisher exact test). These data indicate that a substantial fraction of spontaneous Rad52-YFP foci occur within the rDNA locus in H3 K56R mutant cells.
To further investigate lesions within the rDNA locus in H3 K56R mutants, cells expressing Nop1-CFP were examined for rDNA anaphase bridges, which likely reflect rDNA segregation defects. We determined the percentage of cells with anaphase bridges that coincide with Nop1-CFP. Only 2 to 3% of all H3 WT cells exhibited Nop1-CFP signals stretched between the dividing nuclei during M phase (Fig. 7C). In contrast, 8 to 10% of H3 K56R mutant cells consistently exhibited anaphase bridges in areas that also contained Nop1-CFP, further supporting the notion that RDN1 is a particularly sensitive locus in the absence of H3K56ac. Strikingly, H3K56ac-deficient cells with anaphase bridges marked by Nop1-CFP frequently contained Mre11-, Ddc2-, and Rad52-YFP foci, but this was more rarely observed in H3 WT anaphase cells (Fig. 7C), suggesting that a fraction of H3 K56R mutant cells attempt to segregate chromosomes with damaged rDNA. This is consistent with the data presented in Fig. 4E, showing that H3 K56R mutant cells transiently exposed to MMS segregate DAPI-stained masses of DNA that coincide with Rad52-YFP foci. The observed repair foci and nucleolar anaphase bridges are likely sources of genetic instability. To assay rDNA instability directly, we determined the stability of an ADE2 marker inserted into the rDNA tandem array (55). Interestingly, compared with WT cells, H3 K56R mutants exhibited a 5-fold decrease in marker loss at the rDNA locus (Table 2). This might be consistent with a model in which cells lacking H3K56ac cannot complete an HR-dependent process at the rDNA, thus leading to a reduced frequency of marker loss and to the accumulation of Rad52 foci at RDN1.
In S. cerevisiae, H3K56ac is a modification that occurs in newly synthesized histones deposited behind replication forks throughout the genome. Although a number of genotoxic agents that damage DNA during replication are cytotoxic to cells lacking H3K56ac, the underlying molecular basis of this sensitivity is poorly understood.
We provided evidence that transient exposure to MMS or CPT, but not HU, causes a dramatic loss of viability in cells lacking H3K56ac (Fig. 1, ,4,4, and and5).5). This result is surprising in light of data showing that several replication enzymes seemingly dissociate from replisomes in HU-treated cells that lack H3K56ac (23, 28). Our data argue that loss of replisome components, as monitored by chromatin immunoprecipitation (ChIP) assays, is not sufficient to cause cell death in H3K56ac-deficient cells exposed to HU. In contrast, our DNA combing and PFGE experiments showed that transient exposure to MMS prevented cells devoid of H3K56ac from completing DNA replication (Fig. 2). The presence of permanently damaged replication forks might explain why chromosome XII, which contains the rDNA repeats, is particularly fragile in H3K56ac-deficient cells (Fig. 2F and and7).7). In general, spontaneous or MMS-induced damage to replication forks within RDN1 cannot be rescued by convergent replication (32). This is because a major fraction of the RDN1 locus is replicated in a largely unidirectional manner (62).
Even in WT cells, we found that long stretches of unreplicated DNA remain for several hours after transient exposure to MMS (Fig. 2B). Remarkably, even though rescue of MMS- or CPT-damaged replication forks occurs late in the cell cycle, it depends on the unique chromatin environment created by the deposition of H3K56-acetylated histones behind replication forks. This provides a rationale for the well-established role of DNA damage response kinases in preventing the removal of H3K56ac in response to MMS and CPT (50, 53, 83).
Recent reports have led to the conclusion that H3K56ac plays a role in DNA damage checkpoint deactivation by facilitating chromatin reassembly after extensive DNA end resection (5 kb) of an HO-induced DSB that is repaired by single-strand annealing (SSA) (13, 14). Several of our results argue that this model cannot explain the sensitivity of cells lacking H3K56ac to genotoxic agents such as MMS and CPT. First, transient exposure to 200 mM HU did not cause a significant loss of viability in rtt109Δ or H3 K56R cells, even though HU did activate the DNA damage checkpoint (Fig. 1 and and3B).3B). Second, after transient exposure to MMS, pharmacological inhibition of DNA damage checkpoint kinases did not rescue the loss of viability of H3K56ac-deficient cells (Fig. 3D). Most importantly, many of our results argue that transient exposure to MMS or CPT is cytotoxic not because of a failure to deactivate the DNA damage checkpoint following DNA repair but because they cause persistent DNA lesions in H3K56ac-deficient cells. This interpretation is supported by the fact that segments of unreplicated DNA and Rad52 foci persist for a long time after the removal of MMS or CPT from cells lacking H3K56ac (Fig. 2, ,4,4, and and5).5). Our data do not exclude the possibility that, following extensive DNA end resection and DSB repair by SSA, DNA damage checkpoint deactivation may depend on the restoration of chromatin structure, rather than the completion of DNA repair (13, 14). However, it seems likely that DSB repair between sister chromatids requires only short stretches of DNA resection.
The phenotypes observed in H3K56ac-deficient cells are similar to those of cells lacking the Rtt101, Mms1, or Mms22 protein, which together form a Ub ligase complex involved in the response to DNA damage during replication. Mutation of these genes causes a failure to complete DNA replication, a high incidence of spontaneous Rad52 foci, and a loss of viability after MMS exposure (2, 6, 21, 47, 98). Our survival assays clearly showed that the H3 K56R, rtt101Δ, mms1Δ, or mms22Δ mutation was epistatic to rtt109Δ for MMS sensitivity (Fig. 1E). These results strongly suggest that the functions of the Rtt101-Mms1-Mms22 complex and H3K56ac are interdependent. Interestingly, we found that rtt101Δ and mms1Δ single mutants are considerably less sensitive to MMS than mms22Δ or rtt109Δ mutants (Fig. 1E and data not shown). This suggests that, though part of the same genetic pathway as H3K56ac, Mms22 plays a role in the DNA damage response that is not solely restricted to its function as part of the Rtt101-Mms1-Mms22 Ub ligase complex. Clearly, more experiments are needed to clarify the links between the structure of nascent chromatin and the Rtt101-Mms1-Mms22 Ub ligase, as well as the unique role played by Mms22 in response to MMS.
Gaps and nicks are present behind replication forks stalled by MMS-induced alkylation of DNA bases, but there is also evidence that a fraction of MMS lesions are converted into DNA DSBs (59). This can occur when alkylated DNA bases located ahead of replicated forks are converted into abasic sites by DNA glycosylases and, subsequently, into single-strand nicks by AP endonuclease (59). When replicative DNA polymerases reach these MMS-induced “secondary lesions,” they can, at least potentially, be converted into DNA DSBs. Similarly, the cytotoxicity of CPT results primarily from covalent bonds between DNA topoisomerase I and single-strand nicks located ahead of replisomes. These lesions are converted into DNA DSBs during replication (64). Even though MMS and CPT cause different types of “primary” DNA lesions, both lead to an accumulation of persistent Rad52-YFP foci that last for at least 10 h in a very high fraction of cells lacking H3K56ac (Fig. 4 and and5).5). Remarkably, we found that cells lacking H3K56ac segregate their chromosomes despite the presence of either spontaneous or MMS-induced Rad52 foci (Fig. 4E and and7).7). This is a likely source of MMS sensitivity in H3K56ac-deficient cells, but there may be other sources of lethality as well.
We showed that an rtt109Δ mutation does not prevent error-free DNA template switching (Fig. 6C and D), a Rad51/Rad52-dependent pathway that contributes to the bypass of damaged DNA bases (e.g., 3-methyladenine) that block DNA synthesis by replicative polymerases. Because the cytotoxicity of CPT is largely due to DSBs generated during replication (66), we propose that the persistence of YFP-Rad51 and Rad52-YFP foci following exposure to genotoxic agents is caused, at least in part, by a defect in a homologous recombination-mediated DSB repair pathway that responds to lesions that arise during DNA replication. This argument is consistent with the fact that cells lacking H3K56ac are defective in MMS-induced sister chromatid exchange (21). The step of recombination that may be defective in the absence of H3K56ac is not known. However, our data suggest that it may lie downstream of Rad51 (Fig. 4F), but before the generation of X-shaped structures that correspond to later recombination intermediates (Fig. 6E). At present, we cannot rule out the possibility that, although Rad51 is loaded at sites of DNA damage, it may not be able to perform its function in recombination. One possible scenario is that Rad51 filaments formed in H3K56ac-deficient cells may not be able to promote invasion of the intact sister chromatid. Importantly, several lines of evidence indicate that only a subset of DSB repair events are defective in cells lacking H3K56ac. For instance, H3 K56R cells do not form persistent Rad52 foci after the HO endonuclease generates a DSB at the MAT locus (Fig. 6G to I). In addition, cells lacking H3K56ac efficiently survive DSBs generated by ionizing radiation (40, 53) and are proficient in other forms of recombination-dependent DSB repair (14, 35, 40). Therefore, it is clear that H3K56ac is not generally required for the completion of all forms of recombination-dependent DSB repair.
Interestingly, deletion of RTT109 enhances the MMS and CPT sensitivities of either rad51Δ or rad52Δ mutants (Fig. 6F). To explain these results, the simplest possibility is that a lack of H3K56ac leads to a higher frequency of DSBs than in WT cells, and that this increase in the number of DSBs exacerbates genotoxic-agent sensitivity in cells lacking Rad51 and Rad52. Alternatively, H3K56ac may participate in another, as yet unidentified repair pathway that contributes to MMS and CPT resistance. For example, it seems likely that some of the problems in the completion of DNA replication that we observe in H3K56ac-deficient cells (Fig. 2A to C) might not be associated with DSBs and, consequently, that their resolution in WT cells does not require Rad51 or Rad52. These possibilities are not mutually exclusive, and further studies are needed to elucidate the source of the synergistic MMS and CPT sensitivity observed when rtt109Δ and either rad51Δ or rad52Δ mutations are combined. The acetylation of newly synthesized histones is conserved in human cells. Therefore, the links between nascent chromatin structure and the response to alkylating agents and DNA topoisomerase I inhibitors described here may well apply to human cells, where these genotoxic agents are used in cancer chemotherapy (64, 65, 74).
We thank David Lydall (University of Newcastle), Maria Pia Longhese (University of Milan), Hocine Mankouri, and Ian Hickson (University of Copenhagen) for valuable technical advice.
This work was supported by grants from the Danish Natural Science Research Council (http://en.fi.dk/), the Alfred Benzon Foundation (http://www.benzon-foundation.dk/), and the Villum Kann Rasmussen Foundation (http://www.vkrf.org/vkrf_home.php) to M. Lisby. P. Pasero was funded by the Agence Nationale de la Recherche (http://www.agence-nationale-recherche.fr/) and the Fondation pour la Recherche Médicale (http://www.frm.org/). P. Maddox and A. Verreault were funded by the Canadian Cancer Society Research Institute (FRN-018450) (http://www.cancer.ca/research/) and the Canadian Institutes of Health Research (FRN-79392) (http://www.cihr-irsc.gc.ca/), respectively. The Institute for Research in Immunology and Cancer is supported by infrastructure funds from the Canadian Centre for Excellence in Commercialization and Research, the Canadian Foundation for Innovation, and the Fonds de la Recherche en Santé du Québec.
Published ahead of print 24 October 2011