Genetic Dissection of Budding Yeast PCNA Mutations Responsible for the Regulated Recruitment of Srs2 Helicase

ABSTRACT DNA-damage tolerance (DDT) is a mechanism by which eukaryotes bypass replication-blocking lesions to resume DNA synthesis and maintain cell viability. In Saccharomyces cerevisiae, DDT is mediated by sequential ubiquitination and sumoylation of proliferating cell nuclear antigen (PCNA, encoded by POL30) at the K164 residue. Deletion of RAD5 or RAD18, encoding two ubiquitin ligases required for PCNA ubiquitination, results in severe DNA-damage sensitivity, which can be rescued by inactivation of SRS2 encoding a DNA helicase that inhibits undesired homologous recombination. In this study, we isolated DNA-damage resistant mutants from rad5Δ cells and found that one of them contained a pol30-A171D mutation, which could rescue both rad5Δ and rad18Δ DNA-damage sensitivity in a srs2-dependent and PCNA sumoylation-independent manner. Pol30-A171D abolished physical interaction with Srs2 but not another PCNA-interacting protein Rad30; however, Pol30-A171 is not located in the PCNA-Srs2 interface. The PCNA-Srs2 structure was analyzed to design and create mutations in the complex interface, one of which, pol30-I128A, resulted in phenotypes reminiscent of pol30-A171D. This study allows us to conclude that, unlike other PCNA-binding proteins, Srs2 interacts with PCNA through a partially conserved motif, and the interaction can be strengthened by PCNA sumoylation, which turns Srs2 recruitment into a regulated process.

(suppression of rad-six) mutant was initially isolated by its ability to rescue the severe DNA-damage sensitivity of rad6 and rad18 mutants (10) and subsequently found to encode a DNA helicase (11) that inhibits hyper-recombination (12) by preventing Rad51-ssDNA filament formation (13,14). Srs2 also suppresses the severe DNA-damage sensitivity of other DDT pathway mutants, including rad5 (15)(16)(17). Since the srs2 suppression of DNA-damage sensitivity is limited to only DDT pathway mutants and acts to channel DNA lesions to HR (18), loss of Srs2 is thought to activate a salvage HR pathway (19,20).
PCNA forms a toroidal-shaped homotrimer that interacts with the DNA strand by encircling it, forming a sliding clamp that recruits a large number of proteins to the DNA replication fork not only for replication, but also for other processes, including cell cycle regulation, recombination and DNA damage response (DDR) (21). The most common PCNA-binding motif is known as the PCNA interaction protein (PIP) box, consisting of a consensus sequence Q-x-x-(h)-x-x-(a)-(a) (where "h" represents amino acid with moderately hydrophobic side chains like L, I, M; "a" represents amino acid with highly hydrophobic, aromatic side chains like F, Y; and "x" represents any residues) (22). Several proteins involved in the yeast DDT pathway, including Srs2 (23), Polh (encoded by RAD30) (24,25), Pol2 and all three Pold subunits (26), contain a PIP box or its variants. Correspondingly, analysis of the PCNA-PIP box complex structure (27) reveals that PCNA interdomain connecting loop (IDCL) and C-terminal regions mediate its interaction with the PIP box.
We systematically analyzed DNA-damage resistant isolates from rad5 and rad18 null mutants. While majority of these isolates contained mutations in the SRS2 gene as anticipated, several isolates from rad5D cells did not contain srs2 mutation. Whole-genome sequencing analyses identified putative mutations responsible for DNA-damage tolerance phenotype, among which a missense mutation in POL30 encoding yeast PCNA is further investigated in this study. It was found that PCNA mutations specifically affecting its interaction with Srs2 can activate the salvage HR pathway.

RESULTS
Isolation and characterization of MMS-resistant colonies from rad5D and rad18D cells. Rad18 and Rad5 serve as E3s for PCNA monoubiquitination and subsequent polyubiquitination, respectively (5). In addition, Rad5 is also required for the recruitment of a TLS polymerase Rev1 (28). Hence, Rad18 and Rad5 are required for both branches of DDT, and their null mutants are extremely sensitive to DNA-damaging agents that block replication, including ultraviolet (UV) irradiation, methyl methanesulfonate (MMS), and 4-nitroquinoline N-oxide . It was routinely observed that in an MMS gradient plate assay, some resistant colonies appeared at the high MMS concentration end (Fig. 1A). These resistant colonies turned out to be inherited, as they became significantly more resistant than their parental cells when replating on the MMS gradient plates again (see Fig. 1B for example). It has been well known that srs2 loss-of-function mutations can rescue rad6, rad18 (10) and rad5 (15)(16)(17) severe sensitivity. To ask if these mutants carry srs2 mutations, we sequenced the SRS2 and its promoter region from 33 independent MMS-resistant rad18D isolates and found that they all carried mutations in SRS2 (data not shown). Interestingly, out of 25 independent MMS-resistant rad5D isolates (designated rad5DR), only half (13/25) carried srs2 mutations. Genomic DNA was extracted from the 13 non-SRS2-mutation rad5DR isolates, and the whole-genome sequencing was performed and analyzed. Indeed, these isolates did not contain mutation in SRS2, and the bioinformatics analysis revealed candidate gene mutations, among which a pol30-A171D mutation caught our attention. Fig. 1B shows that a sample rad5DR isolate that contains a srs2 mutation (rad5DR-srs2) is indistinguishable from the reconstituted rad5D srs2D mutant. In comparison, a rad5DR isolate containing a pol30-A171D (designated rad5DR-pol30) mutation is much more resistant than rad5D but less resistant than the rad5D srs2D mutant.
pol30-A171D is synergistic to pol30-K164R in the rescue of rad5D and rad18D sensitivity. To ask whether the pol30-A171D mutation is indeed responsible for the rescue of rad5D cells, we wished to create a pol30-A171D rad5D double mutant. However, POL30 is an essential gene that cannot be readily deleted and replaced by a point mutation. Hence, we took a plasmid shuffling approach as described (29), in which the chromosomal POL30 was deleted while carrying a YCp-Pol30 plasmid with a URA3 selectable marker. Another plasmid carrying a YCp-Pol30 plasmid or its mutant derivatives with a LEU2 selectable marker can be transformed into the above cells and the transformants lost original YCpU-Pol30 can be selected in a 5-FOA plate. The phenotype of a pol30 mutant can then be assessed in comparison to its wild type transformant.
The reconstituted pol30-A171D mutant displayed increased sensitivity to MMS, and it was able to rescue the severe MMS sensitivity of the rad5D mutant ( Fig. 2A). The pol30-K164R mutation was reported to rescue the rad18D sensitivity (5). It was observed that in comparison to pol30-K164R, pol30-A171D cells were less sensitive to DNA-damaging agents both in wild-type and rad5D backgrounds ( Fig. 2A). In addition, pol30-A171D also rescued the rad18D mutants to DNA damage (Fig. 2B).
Since Pol30-K164 is the ubiquitination and sumoylation site (5), it is believed that the pol30-K164R mutation rescues rad18 sensitivity through loss of PCNA sumoylation and compromised Srs2 recruitment (8,9). We first asked whether pol30-A171D affects PCNA sumoylation by a Western blot analysis, which revealed that under the same experimental conditions, pol30-K164R abolished sumoylation at the Pol30-K164 residue while pol30-A171D did not ( Supplementary Fig, S1, lanes 2 and 4). To address genetic relationship between pol30-K164R and pol30-A171D, we created and tested the pol30-K164R,A171D double mutant. To our surprise, the double mutant was much more resistant to all three DNA-damaging agents than its corresponding single mutant, and fully rescued rad5D ( Fig. 2A) and rad18D (Fig. 2B) mutants. In quantitative terms, the pol30-K164R mutation rescued rad5D or rad18D MMS sensitivity by less than 10-fold, and the pol30-A171D mutation rescued rad5D or rad18D MMS sensitivity by more than 10-fold, while the double mutation rescued their MMS sensitivity by as high as 10,000-fold. Hence, pol30-K164R and pol30-A171D are synergistic in rescuing rad5D and rad18D MMS, as well as 4NQO and UV sensitivity.
The rescuing effect of pol30-A171D relies on SRS2. The rescuing effect of pol30-K164R is known to rely on SRS2. To ask whether pol30-A171D also behaves the same, we tested the relative sensitivity of pol30-A171D, pol30-K164R and the corresponding double mutant in a srs2 background. As shown in Fig. 2C, the srs2 mutation is epistatic to all three pol30 mutations with respect to DNA damage sensitivity. The above observations allow us to conclude that pol30-A171D acts in an SRS2-dependent manner.
pol30-A171D affects Pol30-Srs2 interaction. Through the above genetic analysis, we hypothesized that the pol30-A171D mutation affects Srs2 recruitment independently of PCNA sumoylation. To test this hypothesis, A yeast two-hybrid (Y2H) assay was performed to examine physical interaction between Pol30 and Srs2 as well as Siz1. Under our experimental conditions, Pol30 was found to interact with Srs2 but not Siz1 (Fig. 3A). Srs2 was then used to test its interaction with Pol30 mutations using a PIP box-containing protein Rad30 as a positive control. Fig. 3B shows that Pol30-A171D did not affect interaction with Rad30 but severely reduced interaction with Srs2. In contrast, Pol30-K164R did not affect interaction with Srs2 and even enhanced interaction with Rad30. Again, the dual substitution behaved like Pol30-A171D. The Srs2 C-terminal domain (Srs2-CT) was reported to interact with Pol30 and SUMO through its PIP box and SIM motif, respectively (23). In a GST-pulldown assay, Srs2-CT (amino acid residues 1107 to 1174) was cloned into pGEX6 to produce GST-tagged Srs2-CT and His 6 -tagged Pol30 was produced by pET-Pol30 transformed bacterial cells. Under a robust condition of direct interaction between GST-Srs2-CT and His 6 -Pol30, Pol30-K164R did not affect this interaction while Pol30-A171D completely abolished such interaction; in contrast, neither Pol30-K164R nor Pol30-A171D affected physical interaction with the Rad30-PIP domain (Fig. 3C). In conclusion, both Y2H and GST-pulldown assay results support a notion that Pol30-A171D affects Srs2 interaction, but Pol30-K164R does not. It is noted that in both Y2H and pulldown assays, Pol30 was not sumoylated, which explains why Pol30-K164R does not affect interaction with Srs2. Yeast cells cultured overnight were used to make 10-fold serial dilutions and then spotted to YPD or YPD plus various concentrations of MMS or 4NQO or exposed to UV irradiation at indicated doses, followed by incubation at 30°C for 2 days before photography. Only one representative dose for each treatment is shown.

Regulated PCNA-Srs2 Interaction mBio
Distinct effects between pol30-A171D and pol30-K127R. Pol30-K127 can also be sumoylated (5), which regulates PIP-box proteins like EcoI for the sister chromatid cohesion (30). It was previously reported that pol30-K127R and pol30-164R mutations are synergistic in rescuing the rad18 severe DNA-damage sensitivity (8,9,31), implying that sumoylation at the Pol30-K127 residue can backup Pol30-K164 sumoylation for the Srs2 recruitment. It raises a possibility that the Pol30-A171D substitution interferes with Pol30-K127 sumoylation, resulting in the observed mutant phenotypes. Indeed, under our experimental conditions, pol30-K127R itself does not display altered DNA-damage sensitivity or rescuing effects on rad5 and rad18 but causes synergistic phenotypes when combined with pol30-K164R (Fig. S2), reminiscent of the pol30-A171D,K164R double mutant ( Fig. 2A,B). However, pol30-A171D and pol30-K127R are additive in DNA-damage response (Fig. 4A), indicating that they FIG 3 Physical interaction between Pol30-A171D and Srs2. (A) Interaction between Pol30, Srs2 and Siz1 by a Y2H assay. (B) Interaction between mutated Pol30 and Srs2, Rad30 by a Y2H assay. SRS2, SIZ1 and RAD30 ORFs were cloned in pGBT9 (Gal4 BD ), and POL30 ORF and its mutant alleles were cloned in pGAD424 (Gal4 AD ). pGBT9 and pGAD424 derived plasmids were cotransformed into PJ69-4a, and the transformants were spotted on control (SD-Leu-Trp) and selective (SD-Leu-Trp-His with or without 3AT) plates, which were incubated at 30°C for 3 days or as indicated before photography. Independent transformants were examined in parallel on multiple selective plates and incubated for different periods. Only the representative images are shown. (C) In vitro interaction between Srs2 C terminus (Srs2-CT), Rad30-PIP and Pol30 or its mutant derivatives by a GST pulldown assay. Proteins before and after the GST pulldown were subjected to Western blotting by using antibodies against His 6 and GST tags.
Regulated PCNA-Srs2 Interaction mBio function in different pathways. The observed phenotypes were not due to altered cellular PCNA levels, as pol30-A171D, pol30-K164R, pol30-K127R and their corresponding double mutations do not affect protein stability (Fig. 4B). Furthermore, while Pol30-A171D fails to bind Srs2, Pol30-K127R and Pol30-K127R,K164R substitutions do not affect Srs2 interaction ( Fig. 4C), further confirming that pol30-A171D rescues rad5 and rad18 independently of Pol30-K127 sumoylation. Assessment of PCNA IDCL and C-terminal mutations. To ask how the Pol30-A171D substitution affect its interaction with Srs2, we looked at the reported SUMO-PCNA-Srs2-CT complex structure (23). To our surprise, Pol30-A171 is not located in the PCNA-Srs2 interface (Fig. 5A). Srs2-CT has been reported to contain an "atypical PIP box" (23). Comparison of different sources of PIP-box sequences (Fig. 5B) revealed that among conserved residues, Srs2-CT contains an N-terminal "Qxxh" sequence, but not the C-terminal "FF" sequence. In contrast, some other proteins, including Rad30 contain the C-terminal "hxxFF" sequence but not the N-terminal "Q" residue. Since PIP box has been reported to bind two separate regions in PCNA, namely, IDCL (residues 121 to 132) and the C terminus (residues 251 to 258) (27), it is hypothesized that mutations in one of the two PCNA regions result in reduced PCNA-Srs2 interaction and the rescue of rad5D/rad18D mutant phenotypes. To test this hypothesis, two site-specific mutations were created: pol30-2A (L126A, I128A) was to affect the IDCL region and pol30-4A (252-255AAAA) was to affect the C-terminal region (Fig. 5A). In addition, pol30-6A (2A 1 4A) was the combination of the above two mutations (Fig. 5A).
A Y2H assay (Fig. 6A) revealed that Pol30-2A reduced interaction with both Srs2 and Rad30; Pol30-4A affected Rad30 interaction but not Srs2 interaction, while Pol30-6A completely abolished interaction with both proteins. To assess effects of the above mutations on the rescue of rad5D and rad18D severe DNA damage sensitivity, we attempted to shuffle plasmids with these pol30 mutations. To our surprise, only pol30-2A cells were viable, albeit with reduced fitness, whereas pol30-4A and pol30-6A could not replace the plasmid carrying wild-type POL30 with a URA3 selectable marker (Fig. 6B). The above observations indicate that the PCNA C-terminal region plays an essential role, while its IDCL motif is important but dispensable for the cell viability.
The pol30-2A mutation alone caused moderate DNA-damage sensitivity and cannot rescue rad5D and rad18D cells from killing by DNA-damaging agents (Fig. 6C). Hence, a Pol30 mutation affecting interaction with both Srs2 and Rad30 cannot mimic the pol30-A171D phenotype.
pol30-I128A can rescue rad5D and rad18D in a fashion reminiscent of pol30-A171D. To search for a Pol30-A171D mimetic mutation and validate our hypothesis, a group of less destructive mutations were designed either within or around Pol30 IDCL Sidechains of discussed residues are shown in space-filling models with carbon atoms colored in gold. It is noted that in the deposited complex structure, PCNA-127 is a Gly instead of the anticipated Lys. (B) Amino acid sequences of Srs2 PIP box and SIM in comparison to PIP and SIM from other yeast proteins. Consensus sequences in PIP and SIM are highlighted in green and purple, respectively. "h" represents amino acids with moderately hydrophobic side chains like L, I, V and M; "a" represents amino acids with highly hydrophobic, aromatic side chains like F and Y; and "x" represents any residues.
To further address the genetic relationship between pol30-I128A and pol30-K164R, the corresponding double mutation was created, and the two single mutations were found to be strongly synergistic. While each single mutation rescued rad5D or rad18D DNA-damage sensitivity by about 10-fold, the double mutation rescued rad5D and rad18D by at least 1,000-fold under all experimental conditions, to the level indistinguishable from the double mutant alone (Fig. 8A), reminiscent of the pol30-A171D,K164R mutation. As expected, the srs2 mutation is epistatic to pol30-I128A (Fig. 8B). Furthermore, the pol30-A171D,I128A double mutant behaved like the pol30-A171D and pol30-I128A single mutant (Fig. 8C), indicating that they rescued rad5 and rad18 severe sensitivity by the same mechanism. Like pol30-A171D, pol30-I128A also does not affect the PCNA-K164 sumoylation (Fig. S1, lanes 3 and 4). In summary, pol30-I128A mutant phenotypes are similar to pol30-A171D, allowing us to FIG 7 Phenotypes of selected pol30 amino acid substitution mutants. (A) Physical interaction between pol30 mutations and Srs2 or Rad30 by a Y2H assay. Experimental conditions were as described in Fig. 3. (B,C) Phenotypes of pol30 mutants in rescuing rad5D (B) and rad18D (C) mutants from killing by DNA-damaging agents in a serial dilution assay. Experimental conditions were as described in Fig. 2. All plates were incubated at 30°C for 2 days before photography.

DISCUSSION
Previous reports have established that sumoylated PCNA facilitates the Srs2 recruitment, leading to the inhibition of unwanted HR (8,9). This is achieved through coordinated interaction of Srs2-PIP and Srs2-SIM with PCNA and SUMO, respectively, forming a tandem receptor (23). Inactivation of SRS2 can rescue the severe DNA damage sensitivity caused by malfunctional DDT in an HR dependent manner (18), a mechanism known as salvage HR (19,20). Here, we report the isolation and systematic characterization of pol30-A171D that specifically reduces affinity for Srs2 to trigger salvage HR. First, an isolated pol30-A171D mutation can rescue not only rad5D, but also rad18D mutants from killing by DNA-damaging agents, indicating that this mutation induces salvage HR. Second, pol30-A171D does not affect PCNA sumoylationm and is synergistic with pol30-K164R in the rescuing, indicating that they act via distinct mechanisms. Third, since the srs2 mutation is epistatic to pol30-A171D and pol30-K164R, both Pol30 substitutions are speculated to reduce Srs2 recruitment. Finally, pol30-A171D reduces physical interaction with Srs2; pol30-K164R is known to abolish PCNA sumoylation, while the double mutation losses both means to bind Srs2-CT (Fig. 9A), which explains why the two mutations confer synergistic effects in preventing Srs2 recruitment to the replication fork. Furthermore, this study rules out a possibility that pol30-A171D affects the Pol30-K127 sumoylation, leading to the synergistic interaction with pol30-K164R. First, pol30-A171D rescues the severe DNA-damage sensitivity of rad5 and rad18, while pol30-K127R does not. Second, pol30-A171D and pol30-K127R are additive, indicating that they act in different pathways. Finally, Pol30-A171D reduces affinity for Srs2, while Pol30-K127R does not. The above observations support a notion that Pol30-K127 sumoylation can backup Pol30-K164 sumoylation in the recruitment of Srs2 (8,9,23) and hence both functions in the PCNA sumoylation branch (Fig. 9A).
One caveat for the above working model is that, based on the published SUMO-PCNA-Srs2-CT structure (23), Pol30-A171 is not located in the PCNA and Srs2-PIP box interface. To directly test our core hypothesis, a series of rationally designed pol30 mutant derivatives were created and their ability to induce salvage HR as well as genetic interactions with srs2 and pol30-K164R were examined, among which pol30-I128A functionally behaved like pol30-A171D. From the above complex structure, Pol30-I128 is located within the IDCL region in the PCNA-PIP interface, and the Pol30-I128A substitution specifically reduces Srs2 interaction but not Rad30 interaction, providing strong support to our working model (Fig. 9A). One can also see from the above structure that the sidechain of Pol30-A171 is buried in a pocket lined by four sidechains from Pol30-I158, -L151, -L154, -S155, and a b-sheet (Fig. 5A). Based on our collective observations and a speculation that Pol30-A171D also loosens PIP interaction, we predict that amino acid substitution from Pol30-A171 to a larger and hydrophilic Asp residue would destabilize the structure, which can be propagated to the Srs2-binding region(s).
Our Pol30 mutagenesis study reveals that PCNA binds Srs2-PIP and Rad30-PIP differently. First, Pol30-A171D selectively affects Srs2 but not Rad30 binding. Second, disruption of PCNA-IDCL by Pol30-2A reduces its interaction with both Srs2 and Rad30, while disruption of the PCNA C-terminal motif by Pol30-4A selectively interferes with Rad30 but not Srs2 interaction, indicating that the two PCNA regions preferentially bind different consensus sequences in the PIP box. Third, A single amino acid substitution in IDCL by Pol30-I128A has similar effects like Pol30-A171D, while Pol30-F254A in the C terminus only slightly reduces Rad30 binding. Finally, the Pol30-R44A substitution slightly reduces Srs2 interaction and enhances Rad30 binding, reminiscent of Pol30-K164R. The underlying mechanism appears to be that most PIP-boxes adopt a b-stranded structure at the N terminus followed by a 3 10 helix at the C terminus, in which Cdc9-F45 is proximal to Pol30-L126 and I128 (27), whereas Srs2 residues 1153 FSQL 1156 adopt two turns of a-helix instead, in which Srs2-L1156 is proximal to Pol30-I128 and L131 (23). Our observation that Pol30-I128A reduces Srs2 interaction is consistent with a report that Srs2-L1156A also reduces its affinity for Pol30 (23). Apparently, Pol30-I128A would have only a moderate if any effect on the interaction with "FF" motif-containing PIP box proteins. In summary, if a Pol30 mutation only affects Srs2 interaction but not classical PIP box proteins (e.g., A171D and I128A), it could compromise Srs2 recruitment to the replication fork and activate the salvage HR. If a mutation affects both Rad30 and Srs2 (e.g., F254A) or only Rad30 but not Srs2 interaction (e.g., 2A), it cannot rescue the severe DNA-damage sensitivity of DDT mutants. Apparently, interaction with PIP-box proteins is an essential function for PCNA, as disruption of its C-terminal domain cannot maintain cell viability, a conclusion consistent with the structural and biochemical analyses (27). Fortunately, the Srs2-PIP is perhaps unique and hence targeted mutations in PCNA can be made to specifically disrupt its interaction without severe consequences on other PIP-box proteins.
Why have cells evolved such partial PIP boxes to be involved in the interaction with PCNA? One can imagine that proteins containing a full PIP box, including the Cdc9 DNA ligase, Pold , and the Pol« catalytic subunit Pol2, have relatively high affinity for, and constantly travel with, PCNA during replication. On the other hand, a partial PIP box may facilitate regulated association and dissociation from PCNA. For example, Srs2 is efficiently recruited by PCNA only when it is sumoylated (8,9). Polh (Rad30) is recruited to the stalled replication fork when PCNA is monoubiquitinated at the same K164 residue (3) and can be dissociated when PCNA is no longer ubiquitinated (34). Interestingly, in addition to its genome replication role, Pold is also involved in DDT (35 to 37), and some of its subunits contain partial PIP boxes (26). In response to DNA damage that blocks replication, a polymerase switch between Pold (Pol3, Pol31 and Pol32) and Polz 4 (Rev3, Rev7, Pol31 and Pol32) takes place (38,39). Whether this process involves PIP box regulation remains elusive.
Amino acid sequence alignment (Fig. 9B) reveals that PCNA IDCL and C-terminal regions are highly conserved throughout eukaryote organisms, from yeasts to human. The conserved residues include both Pol30-A171 and -I128 (Fig. 9B), suggesting that the regulatory mechanism as described in this study also applies to other eukaryotic organisms, although true Srs2 orthologs have not yet been identified in higher eukaryotes. Hence, this study sheds light on the investigation of DDT pathway regulation in higher eukaryotes, including plants and mammals.
Yeast strains used in this study are listed in Table S1. All strains used in this study were isogenic to HK578-10D except PJ69-4a (40). Gene deletion mutants were created by a one-step gene deletion method (41) using disruption cassettes as described (42), and the target gene deletion was confirmed by genomic PCR, followed by phenotypic analysis. Plasmids were transformed into yeast cells following a LiAc/single-strand carrier DNA/PEG method (43).
Plasmid construction and site-specific mutagenesis. Plasmid YCpL-Pol30 was constructed by cloning the POL30 open reading frame (ORF) along with its own promoter and terminator sequences into YCplac111 (YCp, LEU2) (44) as previously described (45). Site-specific pol30 mutants were created in plasmid YCpL-Pol30 by a modified Quick Change method (46) using primers as shown in Table S2. The resulting plasmids were confirmed by sequencing the entire insert.
For the yeast two-hybrid (Y2H) assay, indicated yeast genes or fragments were amplified by PCR using primers containing restriction enzyme cleavage sites as listed in Table S2, cleaved by the corresponding restriction enzymes and cloned into either pGBT9 or pGAD424 (47) as Gal4 BD or Gal4 AD fusion, respectively. The Escherichia coli transformants were screened by colony PCR and the resulting plasmids were further confirmed by sequencing the entire insert.
To produce recombinant His 6 -tagged Pol30, plasmid pET-Pol30 was constructed by cloning the POL30 ORF into BamHI-SacI sites of pET30a. To produce recombinant GST-tagged Srs2 C-terminal (Srs2-CT) domain, the DNA sequence encoding Srs2 residues 1107 to 1174 was amplified by primers Srs2-CT-BamHI-F and Srs2-CT-EcoRI-R (Table S2) and cloned into the BamHI-EcoRI sites of pGEX6. To produce recombinant GST-tagged Rad30-PIP, DNA sequences encoding Rad30 residues 515 to 632 was amplified by primers Rad30-BamHI-F and Rad30-EcoRI-R (Table S2) and cloned into the BamHI-EcoRI sites of pGEX6. All inserts in the cloned plasmids were confirmed by sequencing.
Plasmid shuffling. A plasmid shuffling method (29) was used to replace plasmid-borne wild-type POL30 with pol30 mutants. Briefly, yeast strain WXY939 and its isogenic mutants contain a chromosomal pol30D::HIS3 allele and the cell viability was maintained by a plasmid pBL211 (YCp, URA3, POL30) (48), a gift from P. Burgers (University of Washington, St. Louis). To shuffle plasmids, WXY939 or its mutant cells were transformed with YCpL-Pol30 derived plasmids, resulting in transformants carrying two plasmids. The transformed cells were cultured in liquid YPD overnight, washed twice with sterile H 2 O, properly diluted and then 0.1 mL culture was spread on a 5-FOA plate. The plate was incubated at 30°C for 3 days and individual colonies were picked and streaked onto a fresh 5-FOA plate. Since the URA3 gene product converts 5-FOA into a compound toxic to yeast cells (49), cells are able to grow only after they have lost the URA3-containing plasmid. Hence, the 5-FOA resistant cells were expected to only carry the YCpL-Pol30 plasmid or its mutant forms, whose phenotypes were then assessed. If the transformed cells do not form colonies on the 5-FOA plate, it is an indication that the pol30 mutation carried by the YCpL-Pol30 plasmid cannot support the POL30 essential function.
Yeast cell survival assays. The gradient plate and serial dilution assays were performed as previously described (50) to assess relative sensitivity of yeast cells to DNA-damaging agents. Briefly, for the gradient plate assay, overnight cultures were printed using a microscopic slide onto a set of two-layer YPD agar plates containing MMS concentration gradients. The plates were incubated for 2 to 3 days at 30°C before photography. For the serial dilution assay, overnight yeast cultures were used to make a set of 10-fold dilutions and then 4.5 mL samples were spotted on freshly made YPD agar plates containing different concentrations of MMS or 4NQO. UV treatment was achieved by exposing yeast cells in the spotted plate to UV irradiation in a UV cross-linker (Stratagene SS-UV1800) at given doses. All plates were incubated for 2 to 3 days at 30°C in the dark before photography. For both assays, only representative and informative plates were presented.
Yeast two-hybrid assay. Plasmids expressing Gal4 AD and Gal4 BD fusion proteins to be tested were cotransformed into yeast strain PJ69-4a. Transformants were allowed to grow at 30°C on an SD-Leu-Trp plate for 2 to 3 days, after which at least two groups of individual colonies were cultured and then spotted on

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.