|Home | About | Journals | Submit | Contact Us | Français|
ATP-dependent chromatin remodeling complexes have been shown to participate in DNA replication in addition to transcription and DNA repair. However, the mechanisms of their involvement in DNA replication remain unclear. Here, we reveal a specific function of the yeast INO80 chromatin remodeling complex in the DNA damage tolerance pathways. Whereas INO80 is necessary for the resumption of replication at forks stalled by methyl methane sulfonate (MMS), it is not required for replication fork collapse after treatment with hydroxyurea (HU). Mechanistically, INO80 regulates DNA damage tolerance during replication through modulation of PCNA (proliferating cell nuclear antigen) ubiquitination and Rad51-mediated processing of recombination intermediates at impeded replication forks. Our findings establish a mechanistic link between INO80 and DNA damage tolerance pathways, indicating that chromatin remodeling is important for accurate DNA replication.
The eukaryotic genome is packaged into chromatin, which limits access to DNA for factors involved in nuclear processes such as DNA replication1. One prominent mechanism of chromatin modification, ATP-dependent remodeling, has been shown to regulate access to the chromatin. Moreover, the INO80 chromatin remodeling complex has been implicated in replication-related activities2–4. The INO80 complex is an evolutionarily conserved ATP-dependent chromatin remodeling complex that was initially described as being involved in transcription5 and DNA repair6–8. However, recent evidence suggests that INO80 is important for the replication of late-firing origins of replication under stress2 as well as for the progression of replication forks after release from HU-induced S-phase arrest3,4. Nonetheless, mechanistic links between chromatin remodeling and DNA replication remain poorly defined1.
We have systematically examined the role of INO80 during replication and found that the INO80 complex is recruited to replication origins (autonomous replicating sequences; ARSs) across the genome under replication stress. Notably, although the ino80 mutant is hypersensitive to HU on plates3–5, we did not identify any important role for INO80 in maintaining replication fork stability in response to HU. However, we discovered that INO80 is specifically involved in the DNA damage tolerance pathway during replication. Here, we show that in the yeast Saccharomyces cerevisiae, INO80 is dispensable for maintaining replication fork stability after HU treatment, but it is required for adequate processing of replication forks in MMS-treated cells through its role in the DNA damage tolerance pathway. Mechanistically, INO80 regulates ubiquitination of PCNA and Rad51-mediated processing of recombination intermediates at impeded replication forks by allowing proper recruitment of Rad18 and Rad51, factors known to be involved in fork resolution. These findings establish INO80 as a newly identified regulator of DNA damage tolerance pathways and reveal the importance of ATP-dependent chromatin remodeling in maintaining genomic integrity during DNA replication.
The hypersensitivities of the ino80 mutant to DNA replication blocking agents such as MMS and HU, together with our observation that Ino80 expression is upregulated specifically during S phase, indicate that INO80 may have a direct role in the response to stalled replication forks5,6. Moreover, the observation that INO80 does not have a major effect on the transcription of known genes implicated in DNA replication (Supplementary Table 1) led us to investigate whether INO80 directly binds ARSs. Given that INO80 was reported to localize to some ARSs in yeast2–4, we extended our analyses to a genome-wide level by combining chromatin immunoprecipitation with high-density oligonucleotide array detection (ChIP-Chip)9. We engineered a Flag-tagged INO80 strain to analyze INO80 localization in cells arrested at the S phase with HU. As a control, we monitored Cdc45 binding to chromosomes to identify early-firing ARSs9. An example of the ChIP-Chip analysis on chromosome VI is shown in Supplementary Figure 1.
Our data indicate that INO80 binds 45% of known ARSs, as defined by the Saccharomyces Genome Database (SGD), during S phase (Fig. 1a). In addition, INO80 binding sites are distributed along all chromosomes (data not shown), suggesting that INO80 is an important factor during replication across the genome. Furthermore, ChIP-Chip analysis of G2 arrested cells (Fig. 1a) shows that INO80 binds only 4% of total ARSs (INO80 expression in G2 and S phase are similar; data not shown), indicating that INO80 binding of ARSs is S-phase specific. Analysis of microarray data originated from S-phase–synchronized, MMS-treated ino80 cells (Supplementary Table 1) indicates that only 6.5% of the INO80 binding signal at ARSs correlate with promoter regions of genes regulated by INO80. By contrast, 45% of INO80 binding signal at promoter regions concur with INO80-regulated genes (see Online Methods for details), suggesting that INO80 activities at ARSs are not related to transcription. Finally, we noticed that INO80 binds both early and late ARSs, with a slight preference for early ARSs (Fig. 1a), indicating that INO80 is broadly required by both early and late ARSs. Together, these findings establish that INO80 is specifically and globally recruited to ARSs during S phase and that the presence of INO80 at ARSs is associated with a replication-related activity.
We first examined whether INO80 is necessary for replication fork movement in unperturbed normal replication using DNA combing analysis, which allows direct visualization of single DNA fibers10. Notably, DNA combing requires a strain constructed in a W303 genetic background that efficiently incorporates bromodeoxyuridine (BrdU). However, INO80 is essential in the W303 background (unpublished observations), so we performed DNA combing using the arp8 mutant instead. Arp8 is an INO80-specific subunit that is required for INO80 ATPase activity, such that the loss of Arp8 mimics the loss of INO8011.
Wild-type and arp8 cultures were incubated in the presence of BrdU to stain newly replicating DNA. DNA was then isolated in plugs and stained with an antibody to BrdU (anti-BrdU)12, and the DNA fibers were quantified under the microscope. Notably, we did not observe any significant difference in BrdU track lengths between wild-type and arp8 cells, indicating that INO80 is dispensable for fork progression during unperturbed replication (Fig. 1b). However, given that ino80 cells are hypersensitive to HU in plates3–6, we investigated whether INO80 is required for fork progression under HU-induced replication stress.
In the presence of HU, the Rad53-dependent intra-S checkpoint is activated through Rad53 phosphorylation13–15. In the absence of Rad53, HU-stalled replication forks collapse and the repression of late firing origins is lost16. As the ino80 mutant is hypersensitive to HU, we investigated whether INO80 is required for fork stability.
Wild-type and ino80 cultures were synchronized in G1 and released into S phase in medium containing 0.2 M HU to arrest replication forks in early S phase. The DNA was then extracted for two-dimensional (2D) gel analysis of ARS305 and ARS1212 (Fig. 1c). As indicated by the arc representing replication fork progression (bubbles), origins were properly fired in both wild-type and ino80 cells. Moreover, HU induced a reversible fork arrest in wild-type cells following G1 release as previously described16, with no signs of fork collapse (ARS305 in Fig. 1c). Unlike the rad53 mutant, ino80 mutant cells were proficient not only in the maintenance of HU-arrested forks, but also in the repression of late-firing origins represented by ARS1212 (fork collapse is marked by the presence of a cone signal in Fig. 1c). Moreover, this effect is global, as we did not observe any difference between wild-type and ino80 cells in the activation of the phosphorylated form of γ-H2AX (a marker of DSB formation17) or the phosphorylated form of Rad53 (a checkpoint activation marker16) during HU arrest or recovery (Fig. 1d). FACS analysis indicated that INO80 is not required for the recovery of replication upon the removal of HU (data not shown). Together, our results suggest that the HU hypersensitivity of the ino80 mutant is not directly related to a defect in replication fork stability or recovery under acute HU-induced replication stress. However, recent studies indicate that mutants of several genes encoding factors involved in the DNA damage tolerance pathways, such as Sgs1 and Rtt101, show no defects in fork stability and recovery after HU treatment when assessed by 2D electrophoresis and pulsed-field gel electrophoresis (PFGE), despite being hypersensitive to HU in plates18,19. As this phenotype resembles the ino80 phenotype, we hypothesized a link between INO80 and the DNA damage tolerance pathway.
In S. cerevisiae, bypass of MMS-induced lesions during S phase is mediated by activation of the RAD6-mediated DNA damage tolerance pathway20–23, and mutants of genes involved in this pathway are hypersensitive to MMS24. Thus, the MMS hypersensitivity of the ino80 mutant5 led us to examine formation of γ-H2AX as a marker of double strand breaks (DSBs) that might result from improper processing of obstructed replication forks17. We found that ino80 mutant cells that were synchronized in G1 and treated with MMS before release into S phase accumulated γ-H2AX only when cells were allowed to progress through the S phase (Fig. 2a). By contrast, cells treated with MMS in G1 that were not released into the S phase did not accumulate γ-H2AX (Supplementary Fig. 2a), indicating that this effect is S-phase specific. Furthermore, the accumulation of Clb2, a G2 cyclin, in the ino80 mutant indicates a delay at the G2 checkpoint due to generation of DSBs during S phase25 (Fig. 2b). These results suggest that INO80 is necessary during DNA replication to avoid DSB generation as a consequence of improper processing of impeded replication forks.
To further investigate whether the accumulation of γ-H2AX in the ino80 mutant is related to deficient replication fork activities, we examined replication fork recovery and progression using PFGE and DNA combing12 (Fig. 2c–g). Wild-type and arp8 cultures released from G1 synchronization were treated with MMS in S phase. MMS was then inactivated and cells were allowed to finish replication.
PFGE analysis showed that, unlike the arp8 mutant, wild-type cells completed S phase within 90 min of MMS inactivation (Fig. 2c,d). Moreover, examination of specific chromosomes by Southern hybridization further confirmed the recovery defects observed in the arp8 mutant (Supplementary Fig. 2d). Consistent with a defect in replication recovery, arp8 mutant cells were arrested in S phase (Supplementary Fig. 2b) and their viability was reduced after MMS treatment (Supplementary Fig. 2c). Moreover, the marked reduction in the length of BrdU tracks (Fig. 2g,e and Supplementary Fig. 2e), together with the persistence of unreplicated gaps observed in the arp8 mutant by combing analysis (Fig. 2f,e and Supplementary Fig. 2e), suggest that there is a defect in fork recovery. Combined, these results establish that INO80 is required to efficiently restart MMS-stalled replication forks and suggest that INO80 is necessary for proper processing of stalled replication forks after MMS treatment.
Given that MMS sensitivity, as well as the inability to resume replication fork movement after MMS treatment in the ino80 mutant, resembles the phenotype of null mutants of genes involved in the DNA damage tolerance pathway21, we investigated whether INO80 might be necessary for activation of the primary step in this pathway, PCNA ubiquitination. After MMS treatment, PCNA is either mono- or polyubiquitinated at its K164 residue, leading to fork resolution through the error-prone and error-free pathways, respectively20. More importantly, a point mutant with a substitution in Lys164 that renders the protein deficient in PCNA ubiquitination (K164R) is also sensitive to MMS26.
To analyze PCNA ubiquitination, wild-type and ino80 cultures were synchronized in G1 and released into S phase in medium containing MMS. Proteins were then extracted in trichloroacetic acid, and PCNA was pulled down under denaturing conditions using anti-PCNA20,27. The ubiquitinated forms of PCNA, although low in abundance, were identified with anti-ubiquitin20 (Fig. 3a). We reproducibly observed that the ubiquitinated forms of PCNA were delayed in induction and moderately but consistently reduced in abundance in the ino80 mutant, whereas expression of the unmodified form of PCNA was not affected. To confirm the identity of the PCNA ubiquitinated bands, we performed the same pulldown analysis using a pcna K164R mutant that cannot be ubiquitinated after MMS treatment26; the PCNA ubiquitinated bands were absent in this strain (Supplementary Fig. 3a). These results indicate that INO80 is necessary for efficient PCNA ubiquitination during replication stress.
The observation that a point mutation (leading to a K737A substitution) that specifically abolishes the ATPase activity of INO80 is also hypersensitive to MMS5 led us to investigate whether INO80 chromatin remodeling activity is required for PCNA ubiquitination. Similar to the ino80 null mutant, the ino80 point mutant strain (K737A) showed reduced PCNA ubiquitination in the same assays (Fig. 3b). To determine whether the delay in PCNA ubiquitination is a consequence of a delay in entering S phase in the ino80 mutant cells, we monitored the acetylation of histone H3 at lysine 56 (H3K56)28, which is specifically modified during S phase and required for replisome stability, as well as the expression of Clb2, another S-phase marker. Both H3 K56 acetylation and Clb2 were induced normally in the ino80 mutant under the same experimental conditions28, indicating that ino80 mutant cells enter the S phase without delay (Supplementary Fig. 3b).
To confirm that the delayed and inefficient PCNA ubiquitination is not an artifact of a delay in entering the S phase in the ino80 mutant, we analyzed PCNA ubiquitination in whole-cell extracts of asynchronous cultures, in which the percentage of S-phase cells is similar in wild-type cells and ino80 mutants. As described, the diubiquitinated form of PCNA increases after MMS treatment in the wild type, but is reduced in both the rad6 and rad5 mutants20. Consistent with the PCNA pulldown experiments, we also observed that PCNA ubiquitination was reduced in the ino80 mutant, whereas both the sumoylated and unmodified forms of PCNA were unaffected (Supplementary Fig. 3c). Similarly, the INO80 ATPase point mutant (K737A) was also defective in PCNA ubiquitination (Supplementary Fig. 3d). Together, these data indicate that INO80 and its chromatin remodeling activities are important for efficient PCNA ubiquitination after MMS-induced replication stress.
The defect in PCNA ubiquitination in the ino80 mutant led us to investigate whether INO80 is necessary to allow recruitment of PCNA ubiquitinating proteins to replication forks. As the first protein known to be recruited to stalled replication forks induced by MMS is the E3 ligase Rad18 (ref. 20), we used ChIP to investigate whether INO80 is necessary for Rad18 recruitment to ARSs (Fig. 3c). Notably, INO80 does not affect the expression of genes involved in the damage tolerance pathway when assessed by genome-wide microarray transcription analysis in MMS-treated, S-phase–synchronized cells (Supplementary Table 1).
Cells tagged at the RAD18 locus were synchronized in G1 and treated with MMS before release into the S phase. Next, cultures were released into the S phase and fork movement was slowed by addition of HU (100 mM) as described23. Our ChIP analysis indicated that, unlike wild-type cells, in which Rad18 signal was enriched at multiple early ARSs (ARS607 and ARS305), Rad18 recruitment is not detectable in the ino80 mutant (Fig. 3c). Furthermore, there was no significant binding signal at a late ARS (ARS 1212) in either the wild type or the ino80 mutant.
As the defect in Rad18 recruitment could be due to either delayed or unsynchronized replication in the ino80 mutant, we performed the same ChIP analysis using Rfc3, a bona fide replisome component, and found that the Rfc3 binding profile was similar in the wild type and the ino80 mutant (Fig. 3d). Next, we monitored the progression of S phase using FACS analysis to show that both wild-type and ino80 mutant cells progress through S phase with synchrony (Supplementary Fig. 3e). Notably, a control primer on chromosome VI indicates that both Rad18 and Rfc3 binding are specific to the ARS sequences (Fig. 3c,d). Finally, the ARS selection for ChIP assays was based on our ChIP-Chip whole genome analysis, and both ARS305 and ARS607 are INO80-binding ARSs, whereas ARS1212 is not. These analyses further establish that even though the normal loading of the replisome is not affected, there is a specific defect in Rad18 recruitment to ARSs in the ino80 mutant. Together, our results indicate that INO80 is required to efficiently recruit Rad18, which initiates the PCNA ubiquitination pathways.
Recent findings suggest that both the RAD6 and RAD51 pathways promote resolution of replication blocks in yeast29. Moreover, Rad51 recombinant activity at obstructed forks generates ‘X-spike structures’, joint DNA molecules that are subsequently resolved by Sgs1. These homologous recombination intermediates are intrinsically different from substrates generated after replication fork collapse or at conventional non–S-phase DSBs30,31. More importantly, Rad51-mediated formation of X-spike molecules at forks requires not only PCNA ubiquitination but also Rad18 activity32. Thus, we investigated whether INO80 is required for Rad51-mediated X-spike generation at obstructed forks (Fig. 4).
Our results confirmed that in the sgs1 mutant, which lacks the helicase required to resolve joint molecules, these structures accumulate and result in enhanced X-spike intensity in 2D gels19 (Fig. 4a, arrow). However, in the rad51 and arp8 mutants these joint molecules are markedly reduced (Fig. 4a,b). Moreover, kinetic studies indicated that the reduction of joint molecules in the arp8 mutant is persistent and is not due to a delay in generating such molecules (Fig. 4c and Supplementary Fig. 4a). These results show that INO80 is involved in the Rad51-dependent processing of replication intermediates after MMS-induced replication block.
As INO80 is necessary to recruit Rad18 to replication forks and to recruit Rad51 during conventional DSB repair33, we examined whether recruitment of Rad51to ARSs also requires INO80. Using anti-Rad51, we performed ChIP analysis under the same experimental protocol used for Rad18 ChIP (Fig. 4d). As previously described23, Rad51 is recruited to ARS607 but not to a control region in wild-type cells during S phase (Fig. 4d). By contrast, recruitment of Rad51 to ARS607 was significantly impaired in the ino80 mutant (Fig. 4d). Furthermore, Rad51 was not recruited to several ARSs that did not show INO80 binding in our ChIP-Chip analysis (ARS501, ARS601 and ARS1212), either in wild-type or in ino80 mutant cells (Supplementary Fig. 4b). These analyses show that INO80 is upstream of Rad51. Together, our results establish INO80 as a regulator of DNA damage tolerance during replication, through its ability to influence the recruitment of factors in both the RAD6 and RAD51 pathways.
Our systematic analysis reveals that INO80 has a specific function in DNA damage tolerance pathways rather than in fork progression or stability (Fig. 4f). Our model suggests that INO80 binds replication forks during S phase and allows access of proteins in the RAD6 and RAD51 pathways to process obstructed replication forks (Supplementary Fig. 5). Chromatin remodeling by INO80 facilitates the recruitment of Rad18 and Rad51 to replication forks, leading to the formation of X-spike structures, a process that requires both Rad18 activity and PCNA ubiquitination. Our study shows that the INO80 chromatin remodeling complex is a relatively early player in this pathway, thus establishing chromatin remodeling as a new regulatory event in DNA damage tolerance during replication.
Our observation that the ino80 mutant grows slower than the wild-type strain indicates that INO80 might have a role during unperturbed DNA replication1,11. However, our combing analysis (Fig. 1b) suggests that INO80 is not necessary during unperturbed replication. Results from our study also indicate that INO80 binds not only to early- but also to late-firing origins across the genome. Although unexpected, our results confirm those of previous work2 and raise the question of whether INO80 has similar or distinct functions at early and late firing origins.
Notably, as the ino80 mutant is hypersensitive to HU, it is plausible that INO80 is involved in replisome stability at stalled replication forks, as previously reported3. However, our analysis and that of others4 show that under HU stress replication forks do not collapse (Fig. 1c). In fact, we did not observe any defect in the recovery of replication upon HU removal, as reported4. Instead, our systematic analyses identified one major and specific role of INO80 in MMS-induced DNA damage tolerance during replication. The differential involvement of INO80 in HU and MMS responses highlights the distinctions among pathways involved in dealing with specific types of replicative stress.
The role of INO80 in damage tolerance is clearly distinct from its role in DSB repair6,8. As previously shown, the nhp10 mutant is defective in DNA resection during conventional DSB repair6, but it is not sensitive to MMS6 or defective in fork recovery after MMS treatment as measured by PFGE (Supplementary Fig. 6a–c). In addition—and similar to the situation with the Rtt101 complex, which is required for fork recovery after MMS but not for homologous recombination at DSBs18,30—Arp8, but not Nhp10, is required for recombination of stalled forks as assessed by the rate of unequal sister chromatid exchange induced by fork stalling at MMS lesions30,34 (Supplementary Fig. 6d). This supports a unique role for INO80 at MMS-stalled forks. Our study has established a previously undescribed role of the INO80 complex in DNA damage tolerance pathways during DNA replication, which is distinct from its role in DSB repair.
Methods and any associated references are available in the online version of the paper at http://www.nature.com/nsmb/.
We thank E. Schwob (IGMM) and J. Rouse (MRC) for strains; H. Ulrich (Cancer UK) for strains and protocols; M.A. Osley (University of New Mexico) for protocols; P. Sung (Yale University) for Rad51 antibody; J. Delrow and R. Basom (Fred Hutchinson Cancer Research Center, FHCRC) for microarray analysis; K. Claypool and S. Hasley (M.D. Anderson Cancer Center (MDACC)) for technical assistance; and members of the Shen lab for comments on the manuscript. This work was supported by funds and grants from MDACC, ACS (RSG-05-060-01-GMC) and the National Institute of Environmental Health Sciences (ES07784) to X.S. (MDACC), from the Rosalie B. Hite Fellowship to K.B.F. (MDACC), from the Fondation Recherche Medicale, the Centre National de la Recherche Scientifique, the Agence Nationale de la Recherche and the Institut National du Cancer to P.P.; by an ARC fellowship to C.A. (CNRS); and by a grant from the US National Institutes of Health (GM71729) to Z.Z. K.S. was supported by a grant of the Cell Innovation Project and Grant-in-Aid for Scientific Research (S) from the Ministry of Education Science and Sports (MEXT), Japan. Y.K. is a Global Center of Excellence (GCOE) research associate.
Accession codes. GEO: MMS-treated INO80 microarray data, GSE18555; INO80 ChIP-Chip dataset, GSE18570.
Note: Supplementary information is available on the Nature Structural & Molecular Biology website.
AUTHOR CONTRIBUTIONS K.B.F. and X.S. designed the study with contributions from X.H., S.W. and Y.S.; K.B.F., C.A., P.P., Y.K., K.S., J.H. and Z.Z. carried out experiments; T.W. and J.X. provided technical assistance. K.B.F. and X.S. wrote the paper.
Reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions/.