In eukaryotic cells, replication past damaged sites in DNA is regulated by the ubiquitination of proliferating cell nuclear antigen (PCNA). Little is known about how this process is affected by chromatin structure. There are two isoforms of the Remodels the Structure of Chromatin (RSC) remodelling complex in yeast. We show that deletion of RSC2 results in a dramatic reduction in the level of PCNA ubiquitination after DNA-damaging treatments, whereas no such effect was observed after deletion of RSC1. Similarly, depletion of the BAF180 component of the corresponding PBAF (Polybromo BRG1 (Brahma-Related Gene 1) Associated Factor) complex in human cells led to a similar reduction in PCNA ubiquitination. Remarkably, we found that depletion of BAF180 resulted after UV-irradiation, in a reduction not only of ubiquitinated PCNA but also of chromatin-associated unmodified PCNA and Rad18 (the E3 ligase that ubiquitinates PCNA). This was accompanied by a modest decrease in fork progression. We propose a model to account for these findings that postulates an involvement of PBAF in repriming of replication downstream from replication forks blocked at sites of DNA damage. In support of this model, chromatin immunoprecipitation data show that the RSC complex in yeast is present in the vicinity of the replication forks, and by extrapolation, this is also likely to be the case for the PBAF complex in human cells.
DNA damage interrupts the progress of replication forks, as replicative DNA polymerases are unable to accommodate most damaged bases in their active sites. In order to deal with this problem, all organisms have evolved a series of specialized DNA polymerases that have lower stringency than replicative polymerases and are able to replicate past different damaged bases (1,2). This is achieved by virtue of their more open active site and the lack of an associated 3′-5′ exonuclease ‘editing function’. Replication past damaged sites—translesion synthesis (TLS)—may take place either at the stalled fork or, following replication restart beyond the lesion, behind the fork (2–4). The sliding clamp accessory protein, β-clamp in Escherichia coli or proliferating cell nuclear antigen (PCNA) in eukaryotes, plays a crucial regulatory role in mediating the switch from the replicative polymerase blocked at a lesion to the specialized TLS polymerase. In eukaryotes, single-stranded regions exposed at the blocked fork trigger the ubiquitination of PCNA on lysine-164 (5,6). This is brought about by the E2 ubiquitin-conjugating enzyme Rad6 and the E3 ubiquitin ligase, Rad18, the latter being activated by single-stranded DNA coated with the single-strand DNA binding protein Replication Protein A (RPA) (7). Most of the TLS polymerases belong to the Y-family (1,8) and all members of this family have ubiquitin-binding motifs, as well as PCNA-binding motifs (9,10) close to their C-termini. Ubiquitination of PCNA thereby, increases the affinity of these polymerases for the PCNA molecules at the blocked forks, providing a mechanism for the switch from replicative to TLS polymerase (6).
Recent work with chicken DT40 cells suggested that TLS at the replication fork was mediated by Rev1, whereas TLS in gaps left behind the replication fork was stimulated by PCNA ubiquitination (11). Further work from the same group revealed that when Rev1 was absent, the transmission of histone modification patterns from parental to daughter nucleosomes was disturbed (12). These results implicated a connection between replication of damaged DNA and the maintenance of chromatin structure.
Chromatin structure is modulated by two classes of proteins. Histone modifiers alter chromatin structure and regulate association of chromatin binding proteins by acetylating, methylating, phosphorylating or ubiquitinating histones (13). Chromatin remodellers use ATP hydrolysis to move, slide or alter the composition of histones within nucleosomes. This is brought about by an ATPase component present in each of the different remodelling complexes (14).
Several reports have implicated chromatin remodelling complexes in responses to DNA damage. For example, both the Remodels the Structure of Chromatin (RSC) and INO80 complexes are recruited to double-strand breaks in yeast (15–19), and in human cells evidence has been provided for a role for INO80 in recruitment of XPC during nucleotide excision repair (20) and for the SWI/SNF complexes (related to budding yeast SWI/SNF and RSC) in phosphorylation of H2AX in response to DNA damage (21–23). INO80 and ISW2 have also been implicated in promoting replication fork progression, particularly at times of replication stress (24–27).
In order to gain further insight into how chromatin structure impacts on replication of damaged DNA, we have investigated the effects of deleting chromatin remodelling genes from Saccharomyces cerevisiae or depleting their mRNAs from human cells. Our data show a major involvement of one of the RSC chromatin remodelling isoforms in ubiquitination of PCNA in yeast and a similar role for the corresponding complex (Polybromo BRG1 (Brahma-Related Gene 1) Associated Factor) or SWI/SNF-B) in human cells.
MATERIALS AND METHODS
The yeast α-PCNA antibody was a kind gift from H. Ulrich. The anti-human polη antibody was raised in rabbit against the full-length protein (28). Other antibodies used in this work are as follows: yeast ubiquitin (P4D1; Cell Signaling Technology), human BAF180 (Bethyl), human PCNA (PC10; Cancer Research, UK), human Rad18 (Abcam), human RPA (RPA70-9; Calbiochem), human histone H3 (Abcam), BrdU (BD) and Myc (9E10; Cancer Research UK).
Yeast strains and plasmids
Yeast cells were cultured in standard YPD (Yeast extract-Peptone-Dextrose) media at 30°C. HisPOL30 and HisPOL30 K164R have been described previously (29,30). rsc1Δ and rsc2Δ were created in the same background. For complementation, pRS413-Rsc1 (JD580) or pRS413-Rsc2 (JD579) plasmids were transformed into rsc2Δ cells and cultured in standard Synthetic Dextrose (SD) (-His) media. For ChIP, a C-terminal 13Myc-tag was introduced in frame with the INO80 or RSC2 gene in BY4741.
Detection of yeast PCNA ubiquitination
Following by treatment with methyl methanesulfonate (MMS), hydroxyurea (HU) or ultraviolet light (UV), HisPCNA was isolated by Ni-NTA under denaturing conditions and detected by western blotting as described previously (7). The dilutions of antibodies were as follows: α-yeast PCNA (1:2500) and α-yeast ubiquitin (1:1000).
Yeast chromatin fractionation
Where indicated, 50 ml mid-log phase cultures were irradiated with 100 Jm−2 and incubated for a further hour at 30°C before being harvested and fractionated. Fractionation involved treatment with zymolyase to create spheroplasts, which were then resuspended in 1 ml lysis buffer (0.4 M sorbitol, 150 mM potassium acetate, 2 mM magnesium acetate, 20 mM HEPES/KOH pH 6.5, 100 µg/ml phenylmethylsulfonyl fluoride, 1 µg/ml pepstatin A, 0.5 µg/ml leupeptin). Spheroplasts were washed three times in lysis buffer, with centrifugation at 4200 rpm for 3 min between each wash. Triton X-100 was added to 1% final concentration and the insoluble fraction was pelleted at 14 000 rpm for 15 min at 4°C. The chromatin pellet was resuspended in 1 x loading buffer and was analysed by western blot.
SV40-transformed MRC5V1, XP12RO (XP-A) and XP30RO (XP-V) fibroblasts were grown in Eagle’s MEM (Invitrogen) supplemented with 2 mM L-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin and 15% foetal calf serum (PAA laboratories).
MRC5V1, XP12RO and XP30RO cells were plated at 5 × 104 cells per 3.5-cm dish and transfected with 20 nM BAF180 siRNA or 20 nM non-targetting control (ON-TARGET plus SMART pool, Dharmacon), and then again with the same siRNAs 24 h later using Hiperfect transfection reagent (Qiagen). UV (10 Jm−2), MMS (1 mM for 1 h followed by 3-h incubation with fresh medium) and HU (1 mM, 24 h) treatments were performed 72 h after the first transfection.
Cells were washed in PBS and Laemmli buffer added directly to the plates. Triton X-100 extraction of cells was carried out as described previously (6). Cells were harvested by scraping and sonicated to shear the DNA and then loaded onto SDS–PAGE gels. After electrophoresis, proteins were transferred to nitrocellulose membranes (GE Healthcare) and immunoblotted with α-BAF180 (1:500), PC10 (1:2500), α-Rad18 (1:2500), α-RPA (1:1000), α-polη (1:1000) or α-histone H3 (1:5000). For quantification, the signal was detected by ImageQuant LAS4000 (GE Healthcare).
Cells were labelled with 40 µM BrdU (SIGMA Aldrich) for 30 min, trypsinized, washed in PBS and fixed in 70% ethanol. Fixed cells were denatured in 2 M HCl–0.5% tritonX-100 for 30 min, then neutralized in 0.1 M NaB4O7 (pH 8.5). Cells were incubated with mouse antibodies to BrdU (1:200) and Fluorescein isothiocyanate -labelled secondary antibodies (1:500), then incubated with 5 µg/ml propidium iodide (Fluka) and analysed in a FACScanto flow cytometer (BD).
Immunolabelling of DNA fibre spreads
Cells were labelled with 20 µM CldU (SIGMA Aldrich) for 20 min and then UV-irradiated (20 Jm−2). Cells were incubated with 200 µM IdU (SIGMA Aldrich) for 20 min. The cells were trypsinized and resuspended in ice-cold PBS at 5 × 105 cells/ml. Cells (2.5 µl) were spotted onto glass slides and lysed with 7.5 µl spread buffer (200 mM Tris–HCl pH 7.5, 50 mM EDTA, 0.5% SDS) for 8 min at room temperature. Slides were tilted by 10–20° for the DNA to run down slowly. They were then air-dried and fixed in 3:1 MeOH/acetic acid. The DNA spreads were rinsed with water and denatured with 2.5 M HCl for 1 h. The slides were washed in PBS, then in blocking buffer (1% BSA, 0.1% Tween20 in PBS) and incubated in blocking buffer for 1 h, then with mouse anti-BrdU (1/500, Becton Dickinson) for 2 h at 37°C, washed in PBS and fixed in 4% paraformaldehyde for 10 min. Next they were washed in PBS and blocking buffer, then incubated with rat anti-BrdU (1/1000, Abcam) overnight at 4°C. After rinsing in 50 mM Tris–HCl pH 8.0, 0.5 M NaCl, 0.5% tween20 for 10 min, slides were washed in PBS and blocking buffer, then incubated with Alexa Fluor 488-labelled chicken anti-rat IgG (1/500, Molecular Probes) and Alexa Fluor 546-labelled goat anti-mouse IgG (1/500, Molecular Probes) for 2 h at 37°C. Finally, the slides were washed in PBS and blocking buffer, then PBS again. The signal was analysed using a zeiss microscope and the fibre signal lengths of 150 replication forks per sample were measured using ImageJ.
Cell survival assays
To measure cell survival, with yeast cells, 5-fold serial dilutions of mid-log phase cultures were spotted onto YPAD (Yeast extract-peptone-adenine-dextrose) media and irradiated with the indicated dose of UV.
With human cells, the MRC5V1 cells were treated with siRNA for 72 h, trypsinized and plated for 6 h before UV irradiation with different doses. Colonies were counted after 10 days of further incubation.
Yeast chromatin immunoprecipitation
Chromatin immunoprecipitation (ChIP) was performed as described (31), with the following modifications: after α-factor arrest, cells were released into 0.2 M HU and samples were fixed with 1% formaldehyde every 20 min. Immunoprecipitation was performed with anti-myc antibody using Protein G Sepharose 4B (Sigma Aldrich). Real-time PCRs were performed using Mx3005P (Stratagene). Primers that amplify regions in the genome correspond to ARS607, the site 4 kb away from ARS607 and a late replicating Tel VI-R region.
RSC and INO80 complexes and PCNA ubiquitination in yeast
The RSC complex is one of the SWI/SNF complex subfamilies. In vitro, it can mediate nucleosome sliding and remove histones from chromatin. The yeast complex contains 17 subunits, some of which are essential (32,33). Strains with deletions of genes encoding non-essential subunits are sensitive to many DNA damaging agents [e.g. Refs (18,34)]. There are two isoforms of the complex, differing only in whether they contain the Rsc1 (10%) or Rsc2 (90%) subunits (35). Rsc1 and Rsc2 have similar structures including two bromodomains and a BAH (Bromo-adjacent homology) domain. Cells in which either subunit is deleted, are viable, whereas deletion of genes encoding both subunits is lethal (35).
We examined the effect of deleting different remodelling genes on PCNA ubiquitination in S. cerevisiae. For these experiments, we used a strain expressing a His-tagged POL30 (PCNA) gene incorporated into the genome as the only source of PCNA (generously provided by H Ulrich) (30). Cells were treated with different agents, incubated for appropriate periods and PCNA was extracted from cell lysates using NiNTA agarose beads. Proteins adhering to the beads were separated on SDS–PAGE gels, which were then immunoblotted with anti-PCNA and anti-ubiquitin antibodies. The former detects unmodified, mono-ubiquitinated and sumoylated PCNA, whereas the latter detects di- and tri-ubiquitinated forms. Treatment of cells with 0.02% MMS for 30 min resulted in robust ubiquitination of PCNA in wild-type cells (Figure 1A, lane 2), but not in a strain in which PCNA was mutated at lysine 164, confirming this as the sole site of ubiquitination of PCNA (Figure 1A, lane 4). Whereas deletion of RSC1 had no effect on PCNA ubiquitination (Figure 1A, lane 6), deletion of RSC2 resulted in a substantial reduction (Figure 1A, lane 8). A similar reduction was found after UV-irradiation (100 Jm−2, 1 h post-UV incubation) (Figure 1B, lanes 2 and 5) or treatment with 100 mM hydroxyurea for 30 min (Figure 1B, lanes 3 and 6). A time course for PCNA ubiquitination after MMS treatment showed substantial reduction in ubiquitination at all times (Figure 1C). The level of unmodified PCNA was not affected by any of the treatments. Overexpression of RSC1 in the rsc2Δ strain failed to restore normal levels of PCNA ubiquitination (Figure 1D, lanes 6 and 8), whereas as expected, expression of RSC2 did restore normal levels (Figure 1D, lanes 10 and 12). We conclude that an intact Rsc2 but not Rsc1 is required for normal levels of PCNA ubiquitination.
Figure 1E shows that rsc2Δ cells were sensitive to UV-irradiation, whereas rsc1Δ cells were not. This is consistent with our findings on the effects of these deletions on PCNA ubiquitination, but we interpret these data with some caution, as deletion of RSC genes may affect other UV repair processes as well.
The INO80 complex is another chromatin remodeller, with 15 subunits in budding yeast (36), that has the ability to replace H2AZ/H2B dimers with H2A/H2B dimers in nucleosomes (37). Consistent with a previous report (27), we found that deleting the gene encoding the Arp8 subunit of INO80 also reproducibly attenuated PCNA ubiquitination in response to MMS treatment (Figure 1F, compare lanes 2 and 6). However, in contrast to the effect of RSC2 deletion, effects seen after UV-irradiation were marginal and not reproducible (Figure 1F, compare lanes 3 and 7) and there was no effect after HU treatment (Figure 1F, lanes 4 and 8) .
RSC complex and PCNA ubiquitination in human cells
There is only a single orthologue of Rsc1/Rsc2 in human cells, a protein-designated BAF180 (or Polybromo), which appears to be a fusion protein of Rsc1, Rsc2 and a third RSC protein, Rsc4. BAF180 contains six bromodomains, followed by two BAH domains and an HMG domain close to the C-terminus (38). BAF180 is a unique subunit of the PBAF (or SWI/SNF-B) remodelling complex, in which BAF180 and two other PBAF-specific subunits (BAF57 and BAF200) are associated with the SWI/SNF core complex (39,40). The PBAF complex affects the transcription of many genes (39,41,42). BAF180 knock-out mice are embryonic lethal (43), and, interestingly, many human cancers have mutations in the PBRM1 gene encoding BAF180 (44,45).
We depleted BAF180 from MRC5 cells using siRNA and measured PCNA ubiquitination by immunoblotting of whole cell lysates with anti-PCNA antibody. The top panels in Figure 2A and B show that the knock-down was effective and the lower panel in Figure 2A shows a substantial reduction of PCNA ubiquitination at different times after UV-irradiation (10 Jm−2), and even in untreated cells. MMS is a relatively weak inducer of PCNA ubiquitination in human cells (Figure 2B, bottom panel), but there was less ubiquitination after treatment with MMS or with HU in cells depleted of BAF180. These results were similar to those of RSC-depleted yeast cells.
In order to eliminate the possibility that the decreased PCNA ubiquitination might be connected in some way with nucleotide excision repair, we repeated the experiments with the nucleotide excision-repair-defective XP-A fibroblasts XP12RO-SV40. The results were very similar to those obtained with MRC5 cells (Figure 2C). Similar results were also obtained with XP30RO cells, defective in DNA polymerase η (Figure 2D). These findings indicate that the deficient PCNA ubiquitination in BAF180-depleted cells is not dependent on NER nor on TLS by polη.
Decreased chromatin association of PCNA and Rad18 in BAF180-depleted cells
In order to explore the possible cause of the decreased ubiquitinated PCNA under conditions of BAF180 depletion, we examined the amounts of chromatin-bound proteins involved in PCNA ubiquitination by immunoblotting extracts of MRC5 cells following extraction with buffer containing 0.2% Triton X-100. As with whole cell extracts, the level of ubiquitinated PCNA in the triton-extracted lysates was much less in BAF180-depleted cells than in the non-targeting siRNA controls (Figure 3A, second panel). Remarkably however, the level of unmodified PCNA in the chromatin fraction was also drastically reduced in the BAF180-depleted samples, though the amount in whole cell lysates was unaffected (e.g. compare with Figure 2A). Furthermore, the amount of chromatin-bound Rad18, the E3 ligase for PCNA ubiquitination, was also markedly reduced (Figure 3A, third panel). Rad18 is recruited to blocked replication forks by virtue of its affinity to RPA-coated single-stranded DNA (7). We therefore measured the chromatin-bound RPA in BAF180-depleted cells, but found only a minor reduction in levels, insufficient to account for the substantial reduction in Rad18. Levels of histone H3 appeared to be unaffected by BAF180 depletion, so, in order to quantitate the effects that we have observed, we expressed the levels of protein of interest, using histone H3 as an internal control and normalized the data to the levels in unirradiated cells in the control samples (Figure 3B). It can be seen that, following UV-irradiation, the amounts of chromatin-bound unmodified (b) and ubiquitinated PCNA (c), Rad18 (d) and RPA (e) all increased substantially after UV-irradiation, albeit to different extents relative to unirradiated cells, and for all except RPA, much of this increase was ablated if BAF180 was depleted. The increase in chromatin-bound RPA was relatively unaffected. We also included polη in several of these experiments and, although there was quite significant variation between experiments, there was no substantial reproducible reduction in chromatin-associated polη [Figure 3B (f)].
We carried out similar analyses following treatment of cells with 1 mM HU for increasing periods of time. The blots are shown in Figure 3C and quantitated in Figure 3D. As with UV treatment, there was an increase in chromatin-bound PCNA and Rad18 upon HU treatment, which was reduced in BAF180-depleted cells. The reduction in unmodified and ubiquitinated PCNA and Rad18 was less than with UV-irradiated cells, but was also observed with RPA.
Using yeast, the data of Figure 1 show that the total amount of PCNA was similar in wild-type and rsc2Δ cells. Figure 3F shows, however, that the level of chromatin-associated PCNA was substantially lower in the rsc2Δ cells.
Cell cycle effects
Loading of PCNA onto chromatin and PCNA ubiquitination are generally DNA replication-associated processes confined to S-phase. We were concerned that the reduced levels of chromatin-bound proteins in BAF180-depleted cells might be the result of a distortion in cell cycle kinetics. Figure 4A shows FACS analysis of the cell cycle distribution of unirradiated cells pulse-labelled with BrdU. The proportion of cells in different phases of the cycle was very similar in the BAF180-depleted and control cells (Figure 4B). Following UV-irradiation, there was, as expected, an increase in the proportion of S-phase cells in both cultures, but this was actually slightly higher in the BAF180-depleted cells (Figure 4B). Thus, we cannot attribute the decreased levels of chromatin-bound proteins to a decreased proportion of S-phase cells.
Failure to ubiquitinate PCNA results in a defect in post-replication repair and substantial UV sensitivity (46). Therefore, to examine the biological effects of the decrease in PCNA ubiquitination brought about by depletion of BAF180, we measured the rate of fork progression using DNA fibre analysis. Cells were labelled for 20 min with chlorodeoxyuridine (CldU), UV-irradiated (0 or 20 Jm−2) and incubated for a further 20 min with iododeoxyuridine (IdU). The distribution of the ratio of CldU to IdU track lengths is shown in Figure 4C (Top panel, unirradiated; bottom panel irradiated). In unirradiated cells, there was no difference in this ratio between the BAF180-depleted cells and those treated with a non-targeting control. In contrast, in UV-irradiated cells, there was a small but reproducible increase in this ratio in the BAF180-depleted cells, compared with the controls. In three experiments the median value of the CldU to IdU ratio was 1.37(±0.20)-fold greater in the BAF180-depleted cells than in the controls. Applying the Kolmogorov–Smirnov test to our data confirmed that the two sets of data were significantly different (P-value of 0.000 for the null hypothesis that the distributions were the same). We conclude that fork progression after DNA damage is less efficient in the absence of BAF180.
Despite the substantial effects described in the preceding sections, the survival of BAF180-depleted and control cells following UV-irradiation was very similar (Figure 4D).
Association of Rsc2 with sites of replication
Ubiquitination of PCNA takes place at the sites of blocked replication forks. Our data implicate a role for Rsc2/BAF180 in this process. Although Rsc2 or BAF180 may remodel chromatin at sites of replication, it is also possible that they are functioning indirectly by facilitating the transcription of genes involved in this pathway. In order to determine whether RSC is present at the replication fork, we have carried out ChIP of myc-tagged Rsc2 in yeast cells synchronized by α-factor and released into HU to maintain them in early S-phase. Figure 5A (left panel) shows some Rsc2 associated with chromatin at three different loci in G1 (compare with untagged cells in middle panel), in keeping with previous reports of chromatin association across the genome. However, when cells were in early S-phase, there was a 3-fold increase in the association of Rsc2 with the early-replicating ARS607 locus (left panel, black and grey bars). The level of association represented an ∼30-fold enrichment over untagged controls. In contrast, the enrichment of Rsc2 at a late-replicating telomere locus did not change significantly over the time course of the experiment (Figure 5A, white bars). The enrichment of Rsc2 was substantially greater than that found with myc-tagged Ino80 (Figure 5A, right panel—see also Refs 24–26) and indicates that the RSC complex is indeed associated with the replication fork.
In this article, we have shown that deletion of the RSC2 gene in yeast or depletion of the orthologous BAF180 in human cells results in a pronounced decrease in the ubiquitination of PCNA following DNA damage or HU treatment relative to undepleted controls. In human cells, we have been able to attribute this reduction to a decreased concentration of the E3 ligase Rad18 in chromatin and remarkably to a decrease in chromatin-associated unmodified PCNA as well. All of these reductions are ∼3- to 5-fold. None of these effects results from distortions of the cell cycle in the absence of BAF180. We can draw two major conclusions from our findings. First, BAF180 and by implication the PBAF complex, plays an important role in loading PCNA onto chromatin and enhancing its ubiquitination following DNA damaging treatment. It is of interest to note that two recent papers report on the role of other chromatin modifications/chromatin proteins on recruitment of PCNA to chromatin in human cells. Mono-methylation of histone H3 on lysine 56 and the HMGN1 protein are both important for recruitment of PCNA to chromatin to facilitate DNA replication (47,48). Our second conclusion is that, whereas the reduced levels of unmodified and ubiquitinated PCNA result in a modest effect on fork progression, they have no effect on UV survival in human cells.
We previously showed that replacement of PCNA in MRC5 cells with mutant PCNA-K164R that cannot be ubiquitinated, resulted in a substantial reduction in UV survival (46). Our current data suggest that, following DNA damage, cells can quite happily dispense with 70–80% of both unmodified and ubiquitinated PCNA with no significant effect on cell survival. (It should be noted that the absence of polη in XP variant cells confers only minor sensitivity to UV-induced cell killing, despite polη being a central player in TLS.) Gong et al. (49) found that a human cell line lacking the Swi–Snf component Brg1 was sensitive to UV-irradiation. However, there are two Swi–Snf complexes in human cells with several components in common, including Brg1. Hence, lack of Brg1 would result in both complexes being defective. In our work, we depleted BAF180, which is a specific component of only one of the complexes, leaving the other one intact. Our data are therefore not directly comparable with those of Gong et al. (49).
A current model for replication at sites of DNA damage is that the PCNA-associated replication machinery is blocked at the site of damage, but the replicative helicase continues to unwind ahead of the fork, exposing single-stranded DNA at the fork. Two events then occur. One of these is the repriming of synthesis downstream of the lesion, using a new PCNA trimer. The other event is the recruitment of Rad18 by the single-stranded DNA to ubiquitinate PCNA. This in turn recruits a TLS polymerase to bypass the damage. Let us first consider the situation in which repriming occurs before TLS (Figure 5B, mode 1). The reprimed replication machinery will be blocked at the next lesion and a further repriming event will take place. Each repriming event requires a new PCNA molecule and each blockage at a lesion results in a new ubiquitination event. If repriming occurs in this way, TLS will take place behind the fork. In an alternative mode (Figure 5B, mode 2), TLS occurs at the fork, the blocked fork is then released to continue with the same PCNA trimer, and repriming may not be necessary. Since this PCNA is already ubiquitinated, another ubiquitination event is not required at the next damaged site encountered. Note that replicative polymerases operate equally well irrespective of whether PCNA is ubiquitinated or not (50), and we have previously shown that PCNA remains ubiquitinated for many hours after UV-irradiation (46).
In this scenario, we propose that PBAF is required to mobilize nucleosomes to allow repriming to take place. This involvement may be either to assist the helicase in unwinding the DNA ahead of the fork or to help the repriming event itself. The latter is perhaps less likely, as repriming takes place on single-stranded DNA, which might be expected to be free of nucleosomes. Several reports have shown that UV-blocked replication forks can restart efficiently by repriming (mode 1) under normal circumstances (51–53). In the absence of PBAF, we suggest that there is a shift in the balance from mode 1 to mode 2, which results in reduced PCNA loading, less recruitment of Rad18 and decreased ubiquitination of PCNA. There is no intrinsic requirement for the involvement of more polη molecules in either mode. If the time interval between repriming and TLS in mode 1 is relatively short, the effect on fork progression need not be very dramatic, and there would not necessarily be any deleterious effect of shifting the balance from mode 1 to mode 2. This would correlate with our finding of a modest decrease in fork progression rate. The model is most easily applicable to events on the leading strand. We note however, that the decrease in chromatin-associated PCNA and ubiquitinated PCNA is >50% when PBAF is depleted, implying a role for PBAF in events on the lagging strand as well. The exact nature of this role must await further experimentation. Our model is highly speculative, but it does provide a satisfactory explanation for our otherwise apparently self-contradictory findings. It posits a role for PBAF/RSC in repriming beyond stalled forks, and predicts that the chromatin remodeller is located at the site of replication. This is consistent with our ChIP data showing that Rsc2 is indeed associated with chromatin in the vicinity of replication forks in yeast cells. In an elegant recent study, Cohen et al. (54) showed that BRG1, a core component of Swi–Snf complexes in mammalian cells, is localized at sites of DNA replication on chromatin fibres and is required for efficient fork progression. This work does not indicate which of the two mammalian Brg1-containing complexes, only one of which (PBAF) contains BAF180, is located and required at the replication forks. We did not find any effect of depleting BAF180 on fork progression in unirradiated cells. However, their results are consistent with our finding that Rsc2 is associated with the replication fork in yeast.
Our data demonstrate that RSC/PBAF has a major effect on PCNA ubiquitination in response to a range of DNA lesions, and that the complex is localized to sites of DNA replication in vivo. We also find, consistent with previous reports, that INO80 facilitates PCNA ubiquitination (55) and is localized to sites of DNA replication. INO80 contributes to PCNA ubiquitination only after MMS-induced DNA damage, but these results still raise the possibility that there are redundant functions of the two chromatin remodellers at stalled replication forks. However, RSC promotes nucleosome repositioning in vivo (19,56), whereas INO80 removes H2A.Z/H2B dimers from nucleosomes and replaces them with H2A/H2B dimers (37). This difference in remodelling activity strongly suggests that the two complexes will promote PCNA ubiquitination in response to MMS by distinct mechanisms. This is consistent with a recent report showing that mutation of the INO80 subunits ies3 or nhp10 together with a rsc4-K25R mutation in RSC results in greater sensitivity to MMS than mutation of either complex alone (57). Importantly, our data suggest that RSC/PBAF impacts on PCNA loading and ubiquitination more broadly in response to DNA damage.
Cancer Research UK [CEA C7905/A10269 to J.A.D.]; the Medical Research Council [Programme Grant G0501450 to A.R.L.]. Funding for open access charge: Research grants.
Conflict of interest statement. None declared.