The 9-1-1 Checkpoint Clamp Physically Interacts with Polζ and Is Partially Required for Spontaneous Polζ-dependent Mutagenesis in Saccharomyces cerevisiae*

The use of translesion synthesis (TLS) polymerases to bypass DNA lesions during replication constitutes an important mechanism to restart blocked/stalled DNA replication forks. Because TLS polymerases generally have low fidelity on undamaged DNA, the cell must regulate the interaction of TLS polymerases with damaged versus undamaged DNA to maintain genome integrity. The Saccharomyces cerevisiae checkpoint proteins Ddc1, Rad17, and Mec3 form a clamp-like structure (the 9-1-1 clamp) that has physical similarity to the homotrimeric sliding clamp proliferating cell nuclear antigen, which interacts with and promotes the processivity of the replicative DNA polymerases. In this work, we demonstrate both an in vivo and in vitro physical interaction between the Mec3 and Ddc1 subunits of the 9-1-1 clamp and the Rev7 subunit of the Polζ TLS polymerase. In addition, we demonstrate that loss of Mec3, Ddc1, or Rad17 results in a decrease in Polζ-dependent spontaneous mutagenesis. These results suggest that, in addition to its checkpoint signaling role, the 9-1-1 clamp may physically regulate Polζ-dependent mutagenesis by controlling the access of Polζ to damaged DNA.

The major replicative DNA polymerases are high fidelity enzymes that can be blocked by lesion-containing bases on the template strand (1). Although such blockage can potentially prevent completion of genome duplication, cells can bypass lesions by copying information from the undamaged sister chromatid in a strand switching or recombination type of mechanism. Alternatively, a translesion synthesis (TLS) 5 DNA polymerase can be recruited to insert a nucleotide across from the lesion or to extend a lesion-base mispair (reviewed in Refs. 2 and 3). This unique activity of the TLS polymerases has been attributed to the presence of a large catalytic active site that can accommodate structurally deformed bases (4). Although it has been suggested that each type of translesion polymerase may be specialized to bypass a certain lesion (or class of lesions) in a relatively error-free manner, many of the TLS polymerases exhibit astoundingly low fidelity on undamaged DNA in vitro (reviewed in Refs. 2 and 4). To minimize replication errors on undamaged DNA, the in vivo use of translesion polymerases must be restricted so that they are employed only when needed.
Saccharomyces cerevisiae contains three translesion polymerases: Pol, Pol, and Rev1. Pol has been primarily characterized with respect to its role in the error-free bypass of UV-induced lesions (5) and this bypass appears to require physical interaction with PCNA (6). Yeast strains lacking REV1 or Pol demonstrate similar phenotypes in response to DNA damage (4), and it is generally assumed that these two polymerases act in the same pathway of mutagenesis (but see Ref. 7). In vitro studies indicate that Rev1 is a G template-specific DNA polymerase (8), but the relevance of this activity to in vivo mutagenesis is unclear (9). Pol, which is comprised of a catalytic subunit encoded by REV3 and a regulatory subunit encoded by REV7 (10), is responsible for at least 90% of UV-induced mutagenesis (11) and 50 -75% of spontaneous mutagenesis (12). Given the highly mutagenic nature of Pol, it is vital for the cell to regulate its interaction with DNA.
Recently, it has been appreciated that many TLS polymerases interact with the processivity clamps that tether the replicative DNA polymerases to the template. Pol IV and Pol V of Escherichia coli, for example, interact with the ␤-clamp homodimer, whereas the eukaryotic TLS polymerases Pol, Pol and Pol physically associate with the PCNA homotrimer (reviewed in Ref. 13). Such interactions have led to the suggestion that a multisubunit processivity clamp might act as a platform during replication to simultaneously tether both replicative and TLS polymerases, thereby allowing ready switching between the two as needed (14). Alternatively, interactions of multiple polymerases with the processivity clamp may be sequential (15), and there is evidence from eukaryotes that post-translational modification of the clamp may be an important regulatory step in polymerase switching (16,17). Interaction of a TLS polymerase with a processivity clamp would not only keep the TLS polymerase in a location to allow the rapid bypass of DNA lesions, but might also sequester a potentially mutagenic TLS polymerase and restrict its access to undamaged DNA.
The Ddc1, Rad17, and Mec3 proteins of S. cerevisiae form a heterotrimeric ring with predicted structural similarity to the PCNA homotrimer (18). This alternative clamp has been termed the 9-1-1 complex based on the names of the homologous Schizosaccharomyes pombe and human proteins (Rad9, Rad1, and Hus1, respectively). The 9-1-1 complex is important in the activation of the DNA damage checkpoint and has been recently found to be associated with several DNA repair fac-* This work was supported in part by National Institutes of Health Grant GM064769 (to S. J.-R.), grants from AIRC, Progetto FIRB-MIUR "Genomica e proteomica nello studio di funzioni cellulari complesse," European Union FP6 Integrated Project DNA Repair (to P. P. and M. M.-F.), and Telethon-Italy Grant GGP030406 (to M. M.-F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 Both authors contributed equally to this work. 2  tors, including the Rad14 nucleotide excision repair (NER) protein of S. cerevisiae (19), the human FEN1 nuclease (20), the human repair DNA polymerase pol ␤ (21), and the MYH glycosylase of S. pombe (22). These diverse interactions may represent a direct function of the 9-1-1 complex in regulating the machineries that process DNA lesions. The 9-1-1 complex can also interact with the DinB TLS polymerase of S. pombe and has been shown to be important for DinB activity during replication stress (23). Finally, genetic evidence suggests that the 9-1-1 clamp may be involved in regulating Pol-dependent damage-induced mutagenesis, as yeast rad17 or mec3 mutants have a reduction in UVinduced mutagenesis similar to that of a rev3 strain (24). The mechanism relating the 9-1-1 complex to Pol activity in induced mutagenesis is not yet understood, and a possible role of this clamp in Pol-dependent spontaneous mutagenesis has not been reported.
In the current study, we identify Rev7 as a partner for 9-1-1 complex components in a two-hybrid screen and we provide evidence that the Mec3 and Ddc1 subunits of the S. cerevisiae 9-1-1 alternative clamp physically interact with Pol. We also show that the 9-1-1 complex is required to stably recruit Pol onto damaged chromosomes, providing an explanation for the role of the complex in UV-induced mutagenesis. Finally, we demonstrate that the 9-1-1 clamp is partially required also for Pol-dependent spontaneous mutagenesis, supporting a model whereby the checkpoint clamp is an important regulator of TLS both under induced and non-induced mutagenesis conditions.
Plasmid and Strain Constructions-Bait plasmids for the yeast twohybrid screening were obtained by amplifying the relevant coding sequences from genomic DNA and ligating the resulting fragments into pEG202 (kind gift from R. Brent). The plasmids used for expressing GST-MEC3 (pFBL7) and GST-DDC1 (pFLB11) in E. coli were derived by inserting appropriate PCR-amplified genomic DNA fragments into pGEX4-T3 (Amersham Biosciences). Specific information regarding these constructs is available on request.
Strains used for the mutation studies were congenic derivatives of SJR922 (MAT␣ ade2-101 oc his3⌬200 ura3⌬Nco lys2⌬A746; Ref. 30). Deletion of RAD1 was as described by Harfe and Jinks-Robertson (31), and a Ura Ϫ derivative was identified on 5-fluoroorotic acid medium to create strain SJR1177. Deletion of REV3, REV7, RAD24, RAD17, DDC1, MEC3, ELG1, and CTF18 was obtained by PCR-mediated gene replacement using pFA6-kanMX2 (28) or pFA6-hphMX4 (32) as appropriate and selecting geneticin-or hygromycin-resistant transformants, respectively. In the rad1 rad24 ctf18 background, ELG1 was replaced with the Kluyveromyces lactis URA3 gene (URA3-Kl), which was amplified using the plasmid pCORE (33) as a template. All gene deletions contained a precise deletion of the published open reading frame, with the exception of REV3, in which the first and last 60 bp of the coding region remain; REV7, in which the first and last 42 bp of the coding region remain; and CTF18, in which 301 bp of the 5Ј and 161 bp of the 3Ј end of the gene remain. Similar disruption strategies of CTF18 have been shown to completely eliminate Ctf18 function (34,35).
In Vitro GST Pull-down Experiments-E. coli BL21 cells transformed with pGEX4-T3, pFLB7, or pFLB11 were grown at 37°C to an A 600 of 0.8. Cells were then shifted to 17°C for 1 h and induced overnight with 1 mM isopropyl 1-thio-␤-D-galactopyranoside at 17°C. Cells were collected, washed with cold H 2 O, and resuspended in cold phosphatebuffered saline containing the Roche Complete protease inhibitor mixture and 1 mM phenylmethylsulfonyl fluoride. Cells were disrupted by sonication, and extracts were clarified by centrifugation at 14,000 ϫ g for 15 min. Extracts were incubated for 2 h with glutathione-Sepharose 4B beads and the resin was then washed 5 times with cold phosphatebuffered saline to remove unbound protein. The resin with bound protein was incubated with 50 l of 35 S-labeled Rev7 for 2 h, followed by extensive washing with phosphate-buffered saline. The 35 S-labeled Rev7 was produced with TNT T7 Quick for PCR DNA (Promega) following the manufacturer's instructions. Proteins that remained bound to the resin were analyzed by autoradiography after SDS-PAGE.
Chromosome Spreading-Cells were grown in YEPD medium at 28°C to a concentration of 5 ϫ 10 6 cells/ml and arrested with ␣-factor (2 g/ml) for 2 h. ␣-Factor was then removed by washing, and cells were allowed to re-enter the cell cycle. Cell cycle progression was assessed by performing FACS analysis at defined times after the release. For UV irradiation, cells were spread on YEPD medium at defined times after the ␣-factor release and irradiated with 100 J/m 2 of UV. Control cells were plated but not treated with UV. Cells were immediately washed off the plates and processed for chromosome spreading as described by Giannattasio et al. (19). Immunofluorescence was performed with 9E10 ␣-Myc, Alexa goat ␣-mouse, and Alexa donkey ␣-goat antibodies. 4,6-Diamidino-2-phenylindole staining was used to identify well spread nuclei, and at least 50 such nuclei were blindly counted for coincident rhodamine staining over 5 slides/experiment for each time point.
Co-immunoprecipitation of Proteins from Yeast-G 1 -arrested cells were released into S phase and UV-irradiated as described above. Cells were immediately washed from plates after UV irradiation and lysed under nondenaturing conditions. Extracts were prepared in phosphatebuffered saline and 10 mg of extract in 1 ml were incubated for 2 h with 7.5 g of either 9E10 ␣-Myc or 12CA5 ␣-HA antibody as appropriate. 30 l of Protein G-agarose beads were added and incubation was continued for another 2 h. Alternatively, extracts were incubated with a resin previously cross-linked to 4G7/11 ␣-Ddc1 antibody (36). After extensive washing, proteins were separated by SDS-PAGE in 10% polyacrylamide gels. Western blotting was performed using standard techniques and ␣-Myc (9E10), ␣-HA (12CA5), or specific antibodies against the protein of interest.
Determination of Spontaneous Reversion Rates and Mutation Spectra-Cultures containing 5 ml of YEPGE were inoculated with single colonies and grown to saturation (2 to 3 days) on a roller drum at 30°C. Cells were pelleted, washed with 5 ml of H 2 O, and resuspended in 1 ml of H 2 O. To assess cell viability and Lys ϩ reversion, appropriate cell dilutions were plated on YEPD and SCD-lysine plates, respectively. Cells were counted after 2 days of growth for YEPD plates and after 2-3 days for SCD-lysine plates. Reversion rates were determined by the method of the median (37), and 95% confidence intervals of the rates were calculated as described in Spell and Jinks-Robertson (38) using Table B11 of Altman (39).
To generate a mutation spectrum, spontaneous Lys ϩ revertants were obtained as described above. To ensure independence of each reversion event, only one colony from each culture was used for analysis. Following purification of revertant colonies, genomic DNA was isolated by glass bead lysis (40) and mutation events were identified by sequencing an appropriate PCR product (see Ref. 30). The reversion rate of a specific category of frameshift event was determined by multiplying the percentage of the event in the reversion spectrum by the corresponding overall reversion rate. To determine the 95% confidence intervals for the event, the high and low confidence intervals for the overall mutation rate were multiplied by the percentage of times the specific event occurred in the spectrum. A table describing the nature of the complex insertions (cins) in each strain is available upon request.

Rev7 Physically Interacts with 9-1-1 Components in Two-hybrid
Experiments-Two-hybrid methods have been extensively utilized to screen for protein interactions in vivo (41), and we have been systematically applying this technique to identify putative interactors with known checkpoint proteins. In the course of a two-hybrid screening using the Ddc1 subunit of the 9-1-1 complex as bait, we identified several REV7-containing plasmids, suggesting that Ddc1 and the Rev7 component of translesion polymerase Pol may interact in vivo. The initial Ddc1-Rev7 interaction identified using a yeast genomic library was confirmed by testing a panel of yeast strains harboring different checkpoint proteins as baits and expressing the Rev7 prey. Cultures of the relevant strains were spotted onto media containing either glucose (prey not expressed) or galactose (prey expressed), and the bait-prey interactions were assessed using lacZ and LEU2 reporters (Fig. 1). Expression of the lacZ reporter indicated a strong interaction between Rev7 and the Ddc1 and Mec3 subunits of the 9-1-1 complex, and these interactions were confirmed by plating cells in the absence of leucine. Interaction of Rev7 with Rad17, the third subunit of the 9-1-1 complex, could not be tested because the Rad17 bait construct partially activated the reporter genes in the absence of any prey (data not shown). The two-hybrid analysis also detected a weak interaction between Rev7 and Rad24, the large subunit of the RFC-like clamp loader that loads the 9-1-1 complex onto DNA in vitro. No interaction was observed, however, when the Rev7 prey was challenged with control baits expressing Drops of yeast cultures were spotted on selective plates containing the indicated carbon source (GLU, glucose; GAL, galactose/raffinose) and incubated for 2 days. Leucine and X-gal were added to the medium as indicated. Mec1-CT and Mec1-NT are bait fusions expressing the C-and N-terminal domains of the Mec1 kinase, respectively. Interaction with a given bait protein results in expression of the lacZ reporter gene, which is detected by the chromogenic substrate X-gal (ϩ LEUCINE/X-GAL) and of the LEU2 reporter gene and growth on plates lacking leucine (Ϫ LEUCINE). Expression of the more sensitive LEU2 reporter gene in the absence of lacZ expression indicates a weak bait-prey interaction. The negative control (Ϫ) corresponds to cells carrying empty bait and prey plasmids. The positive control (ϩ) is a strain carrying p53 and the SV40 T antigen as the bait and prey, respectively. FIGURE 2. Rev7 interacts in vivo with the 9-1-1 clamp and forms a complex with Rev3. Cells expressing physiological levels of the relevant tagged or untagged protein were UV-irradiated and lysed under nondenaturing conditions. Lanes labeled as Ext correspond to the crude extract of the cells expressing the tagged protein. A, protein extracts from cells expressing (ϩ) or not expressing (Ϫ) Rev7-13Myc were incubated with 9E10 anti-Myc antibodies. Proteins recovered using Protein G-agarose beads were analyzed by probing Western blots with Mec3-or Myc-specific antibodies to detect, respectively, Mec3 or Rev7-13Myc. B, protein extracts derived from DDC1 (WT) or ddc1⌬ (⌬) strains expressing Rev7-13Myc were incubated with anti-Ddc1 antibodies crosslinked to protein G-agarose beads. Western blots were probed with Ddc1-and Mycspecific antibodies to detect, respectively, Ddc1 and Rev7. C, protein extracts from Rev7-13Myc cells expressing physiological levels of either HA-tagged Rev3 (ϩ) or untagged Rev3 (Ϫ) were incubated with 12CA5 anti-HA antibodies. Proteins recovered using protein G-agarose beads were analyzed by probing Western blots with HA-specific or Mycspecific antibodies to detect Rev3 or Rev7, respectively. Rev3 was detectable only after overnight exposure. NOVEMBER 18, 2005 • VOLUME 280 • NUMBER 46 the N-or C-terminal halves of Mec1. Altogether, the data obtained with the two-hybrid system indicate that the Rev7 component of Pol interacts with the 9-1-1 clamp, and possibly also with the Rad24-RFC clamp loader.

The Checkpoint Clamp Regulates Translesion DNA Synthesis
Rev7 and 9-1-1 Components Can Be Co-immunoprecipitated from Yeast Extracts-The two-hybrid interaction between Rev7 and the 9-1-1 clamp was confirmed by co-immunoprecipitation experiments using UV-irradiated strains expressing physiological levels of Myctagged Rev7 (Rev7-13Myc). As shown in Fig. 2A, Mec3 was co-immunoprecipitated with Rev7 in extracts derived from the Rev7-13Myc cells, but not from control cells containing untagged Rev7. This interaction was confirmed by the reciprocal experiment (data not shown). Immunoprecipitation with Ddc1-specific antibodies cross-linked to Protein G-agarose beads similarly yielded Rev7-13Myc in the beadbound fractions (Fig. 2B) when wild-type cells were used, but not when the extract was derived from a ddc1⌬ strain.
The cellular level of soluble Rev3 is very low (compare Ext and IP lanes in Fig. 2C). This could explain why we could not co-immunopre-cipitate 9-1-1 components with Rev3-HA, whereas we clearly detected an in vivo interaction between Rev7-13Myc and Rev3-HA (Fig. 2C), in agreement with the finding that Rev7 copurifies with the Rev3 catalytic subunit of Pol (10).
Rev7 Interacts Directly with 9-1-1 Components in Vitro-To obtain further biochemical evidence to support the in vivo interactions of Rev7 with the 9-1-1 components observed by two-hybrid and co-immunoprecipitation analyses, we performed in vitro GST pull-down experiments. For this analysis the REV7 coding sequence was in vitro transcribed and translated in the presence of [ 35 S]methionine. Labeled Rev7 was then incubated with GST-Ddc1 or GST-Mec3 fusion protein purified from E. coli, and polypeptides associated with glutathione-Sepharose beads were analyzed by autoradiography and Coomassie Blue staining after SDS-PAGE (Fig. 3). This analysis demonstrated that labeled Rev7 can indeed interact with both purified Ddc1 and Mec3, suggesting that the observed in vivo interactions are likely to be direct.
The 9-1-1 Complex Influences the Access of Pol to Damaged Chromosomes-It has been demonstrated that the PCNA-like 9-1-1 complex plays an important role in the cellular response to genotoxic stress (42). In this context, it is possible that the 9-1-1 clamp may modulate the recruitment of specific DNA polymerases required for translesion DNA synthesis in response to DNA damage. We thus analyzed the extent of Rev7 binding to chromosomes under normal versus DNAdamaging conditions, and examined whether this loading was influenced by the 9-1-1 complex. Fig. 4 shows the localization of Rev7-13Myc to chromosome spreads. Rev7 (and by inference Pol) clearly bound to chromosomes with increased efficiency after UV treatment, and this preferential binding to damaged DNA was dependent on the presence of a functional 9-1-1 complex.
To confirm this observation and to obtain more quantitative localization data, the chromosome spreads were repeated using synchronous cultures that had been UV irradiated at different times after release from a G 1 arrest. As shown in Fig. 5, the ratio of Rev7-positive nuclei significantly increased in wild-type cells after UV irradiation and this increase was dependent on the presence of Ddc1. The increase in Rev7-positive nuclei was greatest 20 min after release from the ␣-factor G 1 arrest, which corresponded to the time when cells were traversing S phase as assessed by FACS analysis, suggesting that TLS activity may be taking  . Rev7 binds to chromosomes after UV irradiation in a Ddc1-dependent manner. Cells containing Myc-tagged Rev7 were UV irradiated or mock irradiated 20 min after release from an ␣-factor G 1 arrest and chromosome spreads were immediately prepared. The spreads were processed for immunofluorescence with anti-Myc antibodies and DNA was visualized by 4,6-diamidino-2-phenylindole (DAPI) staining. Wild-type (DDC1) and ddc1⌬ strains were compared. place at the blocked replication fork or at gaps left by the replication machinery.
The Alternative Clamp Is Partially Required for Pol-dependent Mutagenesis in Vivo-The physical interaction results agree with previous genetic data suggesting a role for 9-1-1 in UV-induced mutagenesis (24). Moreover, the yeast two-hybrid and GST pull-down experiments indicate that the interaction between Rev7 and Ddc1/Mec3 can occur in the absence of exogenous DNA damage. To determine the biological relevance of this interaction in unperturbed cells, we investigated whether elimination of Rad17, Ddc1, or Mec3 affects spontaneous Pol-dependent mutagenesis. Changes in mutagenesis were assayed using the lys2⌬A746-1 frameshift allele (30). In an NER-deficient rad1 background, two major classes of reversion events are detected: simple frameshifts, which correspond to a single nucleotide insertion, and complex frameshifts, in which an insertion is accompanied by a base pair substitution (31) (Fig. 6). Whereas simple frameshifts are likely the result of random polymerase slippage events that occur during processive replication, complex frameshifts occur primarily in two locations in the reversion window, termed hotspots 1 and 2, and are intimately linked with Pol activity. Complex events are completely absent when Rev3 is eliminated (31) (Fig. 6) or when its catalytic activity is abolished 6 and are believed to reflect lesions that, in the absence of NER, are bypassed in an error-prone manner by Pol (31). Changes in the rate of complex events at hotspots 1 and 2 provide a specific and sensitive indicator of perturbations in Pol-dependent mutagenesis (43).
To determine whether loss of the 9-1-1 PCNA-like clamp affects spontaneous Pol-dependent mutagenesis, we deleted RAD17, MEC3, or DDC1 in a rad1 background. Although the overall lys2⌬A746 reversion rates of the double mutant strains were not significantly different from that of a rad1 strain (TABLE ONE), there was a striking decrease in the rate of accumulation of complex frameshift events at hotspots 1 and 2. Deletion of any one of the subunits of the 9-1-1 alternative clamp resulted in about a 3-4-fold drop in the rate of complex frameshifts at these locations compared with that of a rad1 strain. Examination of the mutation spectrum of a rad1 rev7 mutant demonstrated that Rev7 is required for most, if not all, complex frameshift events in this assay, as no complex events were detected among the 87 revertants sequenced (Fig. 6). These results demonstrate that the alternative clamp is partially required for Pol-dependent mutagenesis in this assay, and validate a biological relevance for the observed physical interaction between Rev7 and Mec3 or Ddc1.

The Rad24 Clamp Loader Protein Does Not Affect Pol-dependent Mutagenesis in the lys2⌬A746 Frameshift Detection Assay-Given that
Pol-dependent translesion synthesis at hotspots 1 and 2 is partially dependent on the 9-1-1 clamp proteins and that Rad24-RFC can load the alternative clamp onto DNA in vitro (18), we predicted that the mutation rate and reversion spectrum of a rad1 rad24 strain should be similar to that of an NER-defective rad17, mec3 or ddc1 strain. As expected, the overall reversion rate of a rad1 rad24 strain did not differ significantly from that of the rad1, rad1 rad17, rad1 mec3, or rad1 ddc1 strains (TABLE ONE). However, complex events at hotspots 1 and 2 did not decrease significantly in the rad1 rad24 mutant relative to the rad1 parent ( Fig. 6 and TABLE ONE). The lack of effect of Rad24 loss on Pol-dependent mutagenesis at hotspots 1 and 2 suggests that the contribution of the 9-1-1 complex to spontaneous mutagenesis is not dependent on the presence of Rad24.
Roles of Ctf18 and Elg1 in Pol-dependent Mutagenesis-Recently, two additional alternative clamp loader-like complexes have been identified in S. cerevisiae and higher eukaryotes: Ctf18-RFC and Elg1-RFC. Like Rad24, Ctf18 and Elg1 form novel complexes with Rfc2-Rfc5 (34, 35, 44 -48) and are speculated to function as clamp loaders. Simultaneous elimination of Ctf18, Elg1, and Rad24 dramatically increases sensitivity of yeast to DNA damaging agents and leads to defects in damage-dependent Rad53 phosphorylation above that seen in the corresponding single or double mutant strains (35, 44 -46). Given the apparent overlapping roles for these alternative clamp loader-like complexes in vivo, we created appropriate triple and quadruple mutant strains to investigate whether loss of two or more of the alternative RFC complexes affects Pol-dependent mutagenesis.
The overall mutation rate in the rad1 rad24 ctf18 strain was ϳ2-fold higher than in a rad1 or a rad1 rad24 strain (TABLE ONE). Unexpectedly, the rate of complex events at hotspots 1 and 2 increased 2-fold in the rad1 rad24 ctf18 triple mutant relative to the rad1 rad24 double mutant strain (TABLE ONE and data not shown). Elimination of Elg1 in a rad1 rad24 ctf18 background increased the overall reversion rate of the lys2⌬A746 frameshift allele nearly 4-fold relative to the triple mutant (TABLE ONE), which is in agreement with other mutation rate studies (44 -46, 49). Although there was a clear decrease in the proportion of complex events at hotspots 1 and 2 compared with other frameshifts in the rad1 rad24 ctf18 elg1 spectrum (Fig. 6), calculation of the 6 S. Yellumahanti and S. Jinks-Robertson, unpublished results. FIGURE 5. Rev7 binding to damaged chromosomes increases during S phase in a Ddc1-dependent manner. DDC1 or ddc1⌬ strains expressing Myc-tagged Rev7 were arrested in G 1 by ␣-factor treatment. After release from the block, samples were UV irradiated or mock irradiated at the indicated times and processed for chromosome spreads. A, FACS analysis of each time sample was done to establish cell cycle position. B, the percentage of Myc-positive spreads at each time point was calculated. The graphs report the ratio of percentages of positive nuclei in the indicated samples. At 20 min after the release from the G 1 arrest, ϳ60% of the nuclei from the UV-irradiated DDC1 strain were Rev7-positive. NOVEMBER 18, 2005 • VOLUME 280 • NUMBER 46 FIGURE 6. Reversion spectra of lys2⌬A746 allele in mutant strains. The total number of independent revertants sequenced for each strain is indicated next to the relevant strain genotype. An insertion of a single nucleotide identical to the surrounding sequence is shown as ϩ. Reversion events in which more than one nucleotide was inserted or events where nucleotides were inserted that were not identical to the surrounding sequence are as indicated below the spectrum. Deletions of two or more nucleotides are shown as Ϫ, complex deletions are shown as cDel, and complex insertions are indicated as cins. Individual complex events that had identical sequence changes are given identical numbers within each spectrum. The number of large deletion events in a given strain is shown boxed above each spectrum. The sequences of hotspots 1 and 2 are indicated by the shaded areas. The rad1 and rad1 rev3 spectra have been published previously (31). rate of complex events at hotspots 1 and 2 in the quadruple mutant revealed no significant decrease relative to a rad1, rad1 rad24, or rad1 rad24 ctf18 strain (TABLE ONE). These data indicate that neither Rad24-RFC, Ctf18-RFC, nor Elg1-RFC is required for efficient Pol-dependent translesion synthesis at hotspots 1 and 2.

The Checkpoint Clamp Regulates Translesion DNA Synthesis
In contrast to the lack of an effect of loss of all three putative alternative clamp loaders on the rate of complex frameshifts, their simultaneous loss was associated with a 16-fold increase in the rate of large deletion events (TABLE ONE). Because such deletions are often elevated in strains with defects in either PCNA or Pol␦ (43,50,51), we suggest that the dramatic increase in large deletions in the rad1 rad24 ctf18 elg1 quadruple mutant may reflect overlapping functions of the corresponding clamp loaders during processive DNA replication.

DISCUSSION
The PCNA-like 9-1-1 complex (composed of Ddc1, Mec3, and Rad17 in budding yeast) is required for the initial steps of the DNA damage signaling cascade that leads to the activation of the checkpoint kinases (52). This complex has been shown to interact both physically and functionally with several DNA repair factors, suggesting a direct involvement in the repair of DNA damage as well (19). To identify new pathways functionally connected to the S. cerevisiae 9-1-1 complex, we performed an extensive two-hybrid screen using the Ddc1 subunit as bait. In the current study, we report that Rev7, the regulatory subunit of Pol, interacts with Ddc1 and Mec3 in two-hybrid assays and in GST pulldown experiments. Furthermore, Rev7 can be co-immunoprecipitated with Ddc1 and with Mec3 from crude extracts obtained from UV-damaged yeast cells expressing endogenous levels of the relevant proteins. These data provide biochemical support for the existence of a complex containing 9-1-1 and Pol. This finding is particularly intriguing because the 9-1-1 complex interacts with the DinB polymerase of S. pombe (23), and its human counterpart stimulates the activity of DNA polymerase ␤ (21).
To investigate the significance of the interaction between 9-1-1 and Pol, we analyzed Pol recruitment to damaged chromosomes. Our results demonstrate that Pol loading increases specifically during S phase following UV irradiation and that this effect requires a fully functional 9-1-1 complex. Deletion of Ddc1 reduced the amount of chromosome-bound Pol to the basal level, eliminating the UV-induced and S phase-specific quota of bound enzyme. One possible interpretation of these data is that Pol may normally be loaded onto chromosomes and, when lesions reach a threshold level, is called into action by the 9-1-1 complex. These results are in agreement with published genetic evidence and may provide a molecular mechanism for the requirement for Rad17 and Mec3 in DNA damage-induced mutagenesis (24). The analysis of 9-1-1 function in Pol activity was extended by investigating the role of the complex in spontaneous mutagenesis. We previously reported that, in the lys2⌬A746 reversion assay, loss of the catalytic subunit of Pol (Rev3) specifically eliminates complex frameshifts that accumulate at two very distinctive hotspots in NER-defective strains (31). Because the Rev3-dependent hotspot events are observed only in the absence of NER, these events are assumed to directly reflect lesion bypass by Pol. The presence/absence of complex frameshifts at these hotspots can thus be used to assess the activity of Pol in spontaneous lesion bypass in vivo (43). The mutation spectrum obtained in an NER-defective rev7⌬ strain was indistinguishable from that obtained in a rev3⌬ background, demonstrating that in this assay both proteins are essential for Pol activity. Further analysis of Pol-dependent complex events revealed that deletion of any one of the genes encoding 9-1-1 subunits resulted in a significant decrease in the rate of complex events at the hotspots, indicating that the interactions between 9-1-1 components and Rev7 are indeed of biological relevance to Pol-dependent activity in spontaneous lesion bypass.
Although the majority of the spontaneous Pol-dependent complex frameshifts required the 9-1-1 components, it should be noted that ϳ30% of these events were independent of the 9-1-1 clamp. This could be a reflection of the context of the DNA lesion. For example, an interaction between Pol and the 9-1-1 clamp may be required if a lesion is

Rates of lys2⌬A746 reversion events in rad1 derivatives
Mutation rates were calculated using at least 14 independent cultures and 95% confidence intervals are indicated in parentheses. Hotspots 1 and 2 seen in the reversion spectra are abbreviated HS1 and 2. Asterisk indicates significant difference relative to a rad1 strain.

Strain
All  NOVEMBER 18, 2005 • VOLUME 280 • NUMBER 46 encountered during leading strand synthesis (as a replication-blocking substrate in the presence of PCNA) as opposed to the lagging strand synthesis, where the lesion might instead reside in a gap. Alternatively, perhaps only certain types of lesions require an interaction between the 9-1-1 clamp and Pol for bypass.

The Checkpoint Clamp Regulates Translesion DNA Synthesis
The S. cerevisiae 9-1-1 heterotrimer is loaded onto DNA in vitro by a modified RFC complex, where Rad24 replaces Rfc1 (18,53,54). We were, therefore, surprised to find that deletion of RAD24 did not affect Pol-specific spontaneous mutation in an NER-defective background. Two additional RFC-like clamp loaders, Ctf18-RFC and Elg1-RFC, have been recently described in S. cerevisiae, and genetic analyses indicate functional overlaps between Rad24, Ctf18, and Elg1 (35, 44 -46). We thus assessed Pol activity in lesion bypass in strains containing deletions of the corresponding genes. Even simultaneous deletion of the ELG1, RAD24, and CTF18 genes did not decrease the rate of complex events at the hotspots. It is possible that either an unidentified clamp loader is involved, or that the replicative RFC clamp loader itself might be important for the activity of the 9-1-1 complex under some circumstances. Intriguingly, a physical interaction between PCNA and the 9-1-1 clamp has been detected both in vivo and in vitro (55,56). 7 Perhaps when the replicative polymerase and PCNA are stalled at a blocking lesion, the 9-1-1 clamp in conjunction with Pol is targeted to (possibly a modified form of) PCNA, thereby allowing Pol to gain access to the DNA lesion for bypass. Finally, an interaction between the Pol32 subunit of the replicative polymerase Pol␦ and the 9-1-1 complex has been suggested (57), and Pol32 has been shown to be required for both spontaneous and induced Pol-dependent mutagenesis (43,50,57,58). Regardless of precisely how the 9-1-1 clamp accesses DNA, our data suggest a Rad24-independent process that, at least in the case of spontaneous damage, is relevant to translesion synthesis and may be separable from the more well characterized checkpoint function.
PCNA has been shown to interact directly with some TLS polymerases and to stimulate their activity in vitro (16, 59 -61). Based on the interaction between Pol and the 9-1-1 clamp reported here, it would be reasonable to predict that 9-1-1 might stimulate Pol. However, recent results failed to detect such a stimulation using purified proteins in an in vitro assay (62). One possible interpretation may be the existence of a yet undescribed accessory factor that is lacking from the biochemically defined system. Our observation, that the 9-1-1 clamp promotes stable binding of Pol to damaged DNA in vivo, suggests that recruitment, rather then direct stimulation of polymerase activity, may be the relevant in vivo function of the 9-1-1 complex. On the other hand, purified PCNA was reported to stimulate Pol activity (62). Because monoubiquitination of PCNA is completely required for induced and partially required for spontaneous Pol-dependent mutagenesis (43,(63)(64)(65), it is tempting to draw direct parallels between the in vitro and in vivo PCNA-Pol results. Caution should be exercised, however, when ascribing in vivo relevance to the stimulatory effect of unmodified PCNA on Pol activity in a purified system that may lack other required accessory factors (e.g. Rev1). As noted by others (13), an alternative role of PCNA monoubiquitination may be to displace a replicative polymerase so that the blocked 3Ј end becomes accessible to the appropriate bypass machinery. The experiments reported here not only add another layer of complexity to lesion bypass by Pol, but also enlarge the role of the 9-1-1 checkpoint clamp during the cellular responses to DNA damage.
How might checkpoints and TLS polymerases cooperate in maintaining genome stability in the absence of exogenous damage? One possibility is that checkpoint factors, including the 9-1-1 clamp, might be associated with the replication fork (66). When a replicative polymerase is blocked by a lesion, checkpoint factors would stabilize the stalled fork; the 9-1-1 clamp might also be involved in preventing further attempts by Pol␦ or Pol⑀ to insert new nucleotides. At the same time, the 9-1-1 clamp, in cooperation with PCNA or Pol32, might drive a polymerase switch to replace replicative enzymes with TLS polymerases. If the lesion is bypassed, replicative polymerases would replace the TLS polymerase to continue DNA synthesis; if not, sufficient single-stranded DNA might be generated by the replicative helicase to trigger a checkpoint cascade. Alternatively the replicative polymerase may restart replication downstream of the lesion, leaving a gap containing the damage, the 9-1-1 clamp, and possibly PCNA behind. This complex would then help recruit a TLS polymerase to the gapped region to complete DNA synthesis. Altogether, the available data strongly suggest that Pol may be considered as a repair component that associates with both 9-1-1 and PCNA clamps. Clearly, a focus of future experiments will be to determine the precise role of the 9-1-1 checkpoint clamp in Pol-dependent mutagenesis, and how the 9-1-1 complex interfaces with PCNA in this process.