Defective DNA replication can result in substantial increases in the level of genome instability. In the yeast Saccharomyces cerevisiae, the pol3-t allele confers a defect in the catalytic subunit of replicative DNA polymerase δ that results in increased rates of mutagenesis, recombination, and chromosome loss, perhaps by increasing the rate of replicative polymerase failure. The translesion polymerases Pol η, Pol ζ, and Rev1 are part of a suite of factors in yeast that can act at sites of replicative polymerase failure. While mutants defective in the translesion polymerases alone displayed few defects, loss of Rev1 was found to suppress the increased rates of spontaneous mutation, recombination, and chromosome loss observed in pol3-t mutants. These results suggest that Rev1 may be involved in facilitating mutagenic and recombinagenic responses to the failure of Pol δ. Genome stability, therefore, may reflect a dynamic relationship between primary and auxiliary DNA polymerases.
THE cell has a large arsenal of mechanisms for preventing genome instability in the form of mutations, genome rearrangements, and loss of heterozygosity (LOH). Efficient DNA replication is critical for normal cellular function, not only because of the necessity to duplicate the genetic information, but also because faulty replication influences the spontaneous frequencies of mutation, genome rearrangement, and LOH arising from nicks, gaps, and breaks in DNA (Horiuchi et al. 1994; Ivessa et al. 2000; Saleh-Gohari et al. 2005). While a variety of DNA repair pathways, including homologous recombination, can provide an efficient and effective means of repairing such DNA damage (Michel et al. 2001; Garg and Burgers 2005a), without the appropriate controls they themselves may lead to increased genome instability (Petes and Hill 1988). These controls are critically important as elevated genome instability can lead to cell death, tumorigenesis, and the development of a range of complex diseases in humans. The normal function of systems involved in DNA replication, recombination, and repair are crucial as they have interdependent responsibilities in maintaining genomic integrity.
DNA replication in yeast is catalyzed by the primary replicative polymerases α, δ, and ε. Pol α synthesizes the primers for leading and lagging strand synthesis, while Pol δ and Pol ε are responsible for the bulk of bidirectional DNA replication (Garg and Burgers 2005b; Johnson and O'Donnell 2005; Pursell et al. 2007; Nick McElhinny et al. 2008). Strains carrying mutations in the POL1, POL2 (CDC17), and POL3 (CDC2) genes, which encode the catalytic subunits of polymerases α (Budd and Campbell 1987), ε (Boulet et al. 1989), and δ (Morrison et al. 1990), respectively, display increased rates of spontaneous mutation and recombination (Aguilera and Klein 1988; Gordenin et al. 1992; Ruskin and Fink 1993; Zou and Rothstein 1997; Kirchner et al. 2000; Pavlov et al. 2001; Galli et al. 2003; Fortune et al. 2005), supporting the link between defective DNA replication and genome instability. In particular, mutations in the POL3 gene that confer a temperature-sensitive growth defect, most likely by affecting the capacity of the cell to replicate its DNA, also confer elevated rates of spontaneous mutation and recombination with a variety of assays (Gordenin et al. 1992, 1993; Tran et al. 1995, 1996, 1997, 1999; Kokoska et al. 1998; Schweitzer and Livingston 1999; Kokoska et al. 2000; Jin et al. 2001; Galli et al. 2003). One of these mutations, pol3-t, is thought to affect the processivity of Pol δ (Gordenin et al. 1992; Tran et al. 1995; Kokoska et al. 2000), which is likely to increase the formation of daughter strand gaps that may be intermediates in the formation of spontaneous mutation and recombination events (Horiuchi et al. 1994; Ivessa et al. 2000; Michel et al. 2001; Minesinger and Jinks-Robertson 2005; Saleh-Gohari et al. 2005; Lopes et al. 2006).
Failure of a replicative polymerase due to an encounter with a spontaneous or induced DNA lesion that blocks its progress provokes a variety of error-free and error-prone responses mediated by a combination of Rad18- and Rad5-dependent post-replication repair and Rad51-dependent recombination repair (Liefshitz et al. 1998; Cejka et al. 2001; Minesinger and Jinks-Robertson 2005). However, in replicative polymerase-defective cells, polymerases may fail without encountering polymerase-blocking lesions, raising the possibility that the processes leading to mutation and recombination may also be different.
The translesion polymerases Pol η, Pol ζ, and Rev1 are recruited to DNA lesions that stall replication forks by blocking advancement of the replicative polymerases (Plosky and Woodgate 2004; Fischaber and Friedberg 2005). Pol η, product of the RAD30 gene (McDonald et al. 1997), possesses the active site plasticity to permit accurate bypass of thymine dimers and 8-oxo guanine lesions (Johnson et al. 1999; Haracska et al. 2000; Prakash et al. 2005), but exhibits high rates of misinsertion at other lesions or undamaged nucleotides (Yuan et al. 2000). Null alleles of RAD30 confer sensitivity to UV light, but no effect on UV-induced mutagenesis, and variable effects on spontaneous mutagenesis (McDonald et al. 1997; Roush et al. 1998), suggesting potential roles in both mutagenic and nonmutagenic lesion bypass mechanisms. Interestingly, Pol η has also been implicated in homologous recombination in chicken cells (Kawamoto et al. 2005), while human Pol η can catalyze DNA synthesis from strand invasion intermediates in vitro (McIlwraith et al. 2005; Rattray and Strathern 2005). Therefore, Pol η may be involved in both mutagenic and recombinagenic responses to stalled replicative polymerases.
Pol ζ is the product of the REV3 and REV7 genes (Morrison et al. 1989; Lawrence and Hinkle 1996) and is required for most spontaneous mutagenesis (Quah et al. 1980; Roche et al. 1994; Kunz et al. 1998; Endo et al. 2007) and for all UV-induced mutagenesis (Lawrence and Christensen 1979; Lawrence and Maher 2001) in yeast. It is also important for seeing the mutations associated with double-strand break (DSB) repair events at the mating-type locus in yeast (Holbeck and Strathern 1997; Rattray et al. 2002) and the immunoglobulin genes in mammals (Diaz et al. 2001; Zan et al. 2001). Perhaps the most relevant biochemical property of Pol ζ is its extraordinary ability to extend from mispaired bases (Prakash et al. 2005; Acharya et al. 2006), which may make it ideal for catalyzing an extension following base insertion opposite a lesion by Pol η. The capacity to drive DNA synthesis from mismatch-containing substrates is likely to be what enables Pol ζ to function during translesion synthesis (Baynton et al. 1998), homologous recombination (Rattray and Strathern 2002; Rattray et al. 2003; Sonoda et al. 2003; Wu et al. 2003), and gross chromosomal rearrangement (Meyer and Bailis 2007). Importantly, a null allele of the REV3 gene was also previously shown to suppress the mutagenic effect of mutations in the POL3 gene, including pol3-t (Pavlov et al. 2001; Northam et al. 2006), suggesting that Pol ζ may be engaged following spontaneous replicative polymerase failure.
Rev1, product of the REV1 gene (Larimer et al. 1989), is a polymerase that is required along with Pol ζ for most spontaneous and induced mutagenesis in yeast (Lawrence 2002); however, its limited deoxycytidyl transferase activity (Nelson et al. 1996) is not required for its function in mutagenesis (Baynton et al. 1999; Haracska et al. 2001). Instead, studies in vitro suggest that Rev1 enhances the capacity of Pol ζ to extend from mismatches and opposite DNA lesions, perhaps through binding to Rev3 (Acharya et al. 2006). Because Rev1 protein levels are 50-fold higher in the G2/M phase of the yeast cell cycle than in the S phase (Waters and Walker 2006), Rev1/Pol ζ-mediated mutagenesis probably occurs at single-stranded regions after the bulk of replication has been completed (Lopes et al. 2006). Restriction of Rev1 activity to the G2/M phase is also consistent with its potential involvement in homologous recombination as suggested by its requirement for gene conversion at immunoglobulin gene loci in chicken cells (Okada et al. 2005). Interestingly, the REV3 and REV7 genes were not required for these events, consistent with the suggestion that Rev1 can participate in Pol ζ-dependent and -independent events (Baynton et al. 1999; Okada et al. 2005).
The work presented here explores the responses by the translesion polymerases Pol η, Pol ζ, and Rev1 to defective polymerase δ in yeast strains bearing the pol3-t mutation. We observed that, while the pol3-t mutation conferred significantly elevated rates of mutation, recombination, and chromosome loss, null alleles of the RAD30, REV1, REV3, and REV7 genes alone had few effects. However, combining the pol3-t allele with the translesion polymerase mutations revealed that loss of REV1 consistently suppressed the elevated rates of mutation, recombination, and chromosome loss conferred by pol3-t. These results are consistent with Rev1 responding to replicative polymerase failure in pol3-t mutant cells by eliciting a broad spectrum of genome-destabilizing events, perhaps by facilitating the interaction of defective Pol δ with daughter strand nicks or gaps.
MATERIALS AND METHODS
Yeast strains, plasmids, and growth conditions:
All of the yeast strains used in this study are isogenic with W303-1A (Thomas and Rothstein 1989) and derived from the strains listed in Table 1. All strains used in this study contain the wild-type RAD5 allele. Standard methods were used for the construction, growth, and maintenance of yeast strains (Burke et al. 2000). Isolation of the pol3-t mutant allele has been previously described (Kokoska et al. 1998). The pol3-t allele was incorporated into the W303 strain background by pop-in–pop-out (Rothstein 1991), using the plasmid p171, the generous gift of Dmitri Gordenin, and was maintained in a heterozygous state in diploid strains. Since the pol3-t mutation is believed to confer rapid genome destabilization, producing secondary mutations shortly after germination, all strains containing a pol3-t allele used in our experiments were derived from spore colonies taken directly from dissection plates that had been maintained at 23° for no longer than 3 days. Segregants containing pol3-t were identified by their temperature-sensitive growth at 37°. The rad30∷HIS3, rev1∷HIS3, rev3∷hisG-URA3-hisG, and rev7∷hisG-URA3-hisG alleles were crossed into our laboratory strains using W303-derived strains that were the generous gift of John McDonald and Roger Woodgate.
The rad30∷LEU2 allele was generated by single-step gene disruption (Rothstein 1991) using a construct generated in vitro as described below. Primers P1 (5′-CCT TAT CGC GGC GAA AAA AGC GAC GGT CGA GGA GAA CT C-3′) and P2 (5′-GGT ACT TCG TTC TTC TTA TCG GTT CAA GAA GGT ATT GAC-3′) were used to clone the LEU2 gene from plasmid pRS415 (Sikorski and Hieter 1989), producing fragments with ends consisting of 18 bp of homology to the genomic sequences immediately flanking the site of HIS3 marker insertion in the rad30∷HIS3 allele. Primers P3 (5′-CCT GCC GAT CAT AGG ATA CC-3′) and P4 (5′-CTT TTT TCG CCG CGA TAA GG-3′) and primers P5 (5′-GAT AAG AAG AAC GAA GTA CC-3′) and P6 (5′-GAC TTC CAA ATC TCT ATC-3′) were used to clone 155- and 138-bp fragments homologous to sequences upstream and downstream from rad30∷HIS3, respectively. These fragments each share homology with one end of the fragment produced from pRS415. The three PCR-generated fragments were then used as templates for primers P3 and P6 to produce a single rad30∷LEU2 fragment that was then integrated into the genome using lithium acetate transformation (Schiestl and Gietz 1989; Manthey et al. 2004). Segregation against the rad30∷HIS3 allele in genetic crosses and Southern blot analyses (data not shown) were carried out to verify insertion of the rad30∷LEU2 construct into the RAD30 locus.
The rev1∷KAN-MX allele was generated by single-step gene disruption as described below. Primers REV1-F3943 (5′-CAA TTC CCA GCT CGT CCC-3′) and REV1R-6530 (5′-GCT CAC TGT GCA ACC ATT CG-3′) were used to amplify a 2587-bp DNA fragment carrying the wild-type REV1 sequence from genomic DNA. The ends of the fragment were made blunt with T4 DNA polymerase and cloned into pBlueScript (Stratagene) that had been digested with HincII to create the plasmid pLAY568. pLAY568 was digested with HincII to remove 797 bp of DNA encompassing 53 bp of DNA 5′ to the initiation codon for REV1 and 741 bp downstream. A 1483-bp DNA fragment containing the KAN-MX selectable marker generated by SmaI and EcoRV digestion of the plasmid pFA6-KAN-MX was inserted into HincII-digested pLAY568 to generate pLAY571. Digestion of pLAY571 with XbaI and XhoI released a 3290-bp rev1∷KAN-MX fragment that was electroporated into yeast, followed by selection for resistance to G418. The structure of the disrupted REV1 locus was confirmed by Southern blot analysis and segregation against the rev1∷HIS3 allele in genetic crosses (data not shown).
Determination of spontaneous mutation rates:
Spore colonies were excised from plates containing freshly dissected tetrads incubated at 23° for 2–3 days and dispersed in dH2O. For the CAN1 mutation assay, aliquots of cell suspension were plated on synthetic medium lacking arginine and supplemented with 60 μg/ml canavanine and incubated for 4 days at 30°. For the hom3-10 reversion assay, aliquots of cell suspension were plated on synthetic medium lacking threonine and incubated for 4 days at 30°. Viable counts were determined by plating appropriate dilutions of cell suspension onto synthetic complete medium and incubating for 4 days at 30°. Mutation rates were determined by the method of the median (Lea and Coulson 1949). Confidence intervals were determined as previously described (Spell and Jinks-Robertson 2004). Statistical significance was evaluated using the Mann–Whitney test.
Determination of mutation spectrum by DNA sequence analysis:
Single canavanine-resistant colonies were selected from 48 independent cultures of each genotype and genomic DNA was prepared by glass bead disruption and phenol:chloroform extraction. Sequences encompassing the CAN1 gene and its promoter were amplified from each sample by PCR using the primer pairs 298D (5′-TTT CGA GGA AGA CGA TAA GGT-3′) and 803U (5′-GCA CCT GGG TTT CTC CAA T-3′) and 679D (5′-GAG TTC TGG GTC GCT TCC ATC-3′) and 1841U (5′-GTA TGA CTT ATG AGG GTG AGA-3′). Nucleotide sequences were determined by automated fluorescence sequencing using the primers 276D (5′-TAT TGG TAT GAT TGC CCT TG-3′), 404U (5′-GAA TAT GCC AAA GAA CCC-3′), 679D (5′-GAG TTC TGG GTC GCT TCC ATC-3′), and 1150D (5′-ACA ACC ATT ATT TCT GCC GC-3′). Mutations were confirmed by reamplifying and sequencing in the opposite direction. Statistical significance of the differences in mutation spectrum was evaluated using Fisher's exact test and contingency chi-square analysis.
Determination of spontaneous unequal sister-chromatid recombination rates:
Spontaneous unequal sister-chromatid recombination was assayed as previously described (Fasullo and Davis 1987). Briefly, haploid strains that carry a trp1-1-linked direct repeat of 5′- and 3′-deleted his3 sequences, arranged tail-to-head around a URA3 marker, were used to measure interchromatid recombination. Spore colonies were obtained from freshly dissected tetrads that had been incubated at 23° for 2–3 days and dispersed in dH2O. Aliquots of suspended cells were plated on synthetic medium lacking histidine and allowed to grow at 30° for 4 days. Viable counts were determined by plating appropriate dilutions onto synthetic complete medium and incubating at 30° for 4 days. Rates of sister-chromatid recombination were determined by the method of the median (Lea and Coulson 1949). Confidence intervals were determined as previously described (Spell and Jinks-Robertson 2004). Statistical significance was evaluated using the Mann–Whitney test.
Determination of intrachromosomal recombination rates:
Spontaneous intrachromosomal recombination was assayed using a construction that was previously described (Maines et al. 1998). Spore colonies carrying 3′- and 5′-deleted his3 segments that share 415 bp of HIS3 coding sequence flanking a URA3 marker at the HIS3 locus were dispersed in dH2O. Aliquots of suspended cells were plated on synthetic medium lacking histidine and incubated at 30° for 4 days to select for recombinants with a complete HIS3 allele. Viable counts were determined by plating appropriate dilutions on synthetic complete medium and incubating at 30° for 4 days. Recombination rates, confidence intervals, and statistical significance were determined as described above.
Determination of spontaneous chromosome loss and interhomolog recombination rates:
Spontaneous loss of chromosome V and interhomolog recombination were assayed as previously described (Klein 2001). Briefly, individual zygotes were micromanipulated onto selective medium to ensure diploidy and allowed to grow at 23° for 3–4 days. Colonies were excised from the plates and dispersed in dH2O. Aliquots of suspended cells were plated on synthetic medium lacking arginine and supplemented with 60 μg/ml canavanine and incubated at 30° for 3–4 days to determine the number of canavanine-resistant cells. Canavanine-resistant colonies were replica plated to synthetic medium lacking threonine, and the replicas were incubated at 30° for 2 days to determine the fractions of colonies that had acquired their canavanine resistance through interhomolog recombination (CanR Thr+) or loss of chromosome V (CanR Thr−). Viable counts were determined by survival on synthetic complete medium after incubation at 30° for 3–4 days. Recombination and chromosome loss rates, confidence intervals, and statistical significance were determined as described above.
Elevated mutation rates in the pol3-t mutant are suppressed by rev1Δ:
Several studies have documented significantly increased rates of mutation in pol3-t mutant strains with a variety of assays (Gordenin et al. 1992; Tran et al. 1995, 1996; Gordenin and Resnick 1998; Kokoska et al. 1998; Galli et al. 2003). We observed an ∼10-fold increase in the rate of mutation of the CAN1 gene (Table 2), indicative of a general mutator effect (Whelan et al. 1979). The pol3-t allele had only a 2.5-fold effect on the rate of reversion of the hom3-10 allele (P = 0.001), a measure of the propensity toward frameshift mutation (Flury et al. 1976; Marsischky et al. 1996). The modest effect of the pol3-t allele on frameshifting in our assays suggests that its effect on the general mutation rate may not be primarily due to slippage of Pol δ during DNA synthesis (Tran et al. 1996).
Loss of the translesion polymerases Rev1 and ζ themselves had no significant effect on mutation of CAN1 or hom3-10, as the mutation rates in the rev1Δ (P = 0.8 or 0.2), rev3Δ (P = 0.6 or 0.2), and rev7Δ (P = 0.4 or 0.3) single mutants were not significantly different from wild type (Table 2). Loss of Pol η had a slight but significant effect on mutation of CAN1 (P = 0.009) but no significant effect on the reversion of hom3-10 (P = 0.5). These results suggest that these polymerases have minimal individual impact on spontaneous mutagenesis in our strains. When combined with pol3-t, however, loss of Rev1 completely suppressed the elevated rate of CAN1 mutation conferred by pol3-t, as the mutation rate in the pol3-t rev1Δ double mutant was not significantly different (P = 0.17) from that in wild type, indicating that Rev1 facilitates CAN1 mutagenesis in the presence of a defective Pol δ. In contrast, no significant effect of Rev1 was observed on hom3-10 reversion in pol3-t mutants, as the rate in the pol3-t rev1Δ double mutant was not significantly different (P = 0.84) from that in the pol3-t single mutant, suggesting that Rev1 does not facilitate replicative polymerase slippage. Interestingly, the potent suppressive effect of the rev1Δ allele on stimulation of CAN1 mutation by pol3-t was not observed for the rad30Δ, rev3Δ, or rev7Δ alleles, as the rates in the pol3-t rad30Δ (P = 0.98), pol3-t rev3Δ (P = 0.72), and pol3-t rev7Δ (P = 0.79) double mutants were not significantly different from those in pol3-t single-mutant cells. This runs counter to the results of previous studies that indicated that rev3Δ can suppress the effects of pol3 alleles, including pol3-t (Pavlov et al. 2001; Northam et al. 2006), on mutagenesis of CAN1, suggesting that Pol ζ may exert different effects on mutagenesis in different yeast strains. The rad30Δ, rev3Δ, and rev7Δ alleles, however, do exert an effect on reversion of hom3-10 in pol3-t mutant cells, as the rates are four- to fivefold higher in the pol3-t rad30Δ, pol3-t rev3Δ, and pol3-t rev7Δ double mutants than in the pol3-t single mutants. This suggests that Pol η and Pol ζ may suppress slippage of Pol δ in pol3-t mutant cells or may promote repair responses that oppose other mechanisms of frameshift formation (Tran et al. 1996).
The pol3-t mutation confers a distinct mutation spectrum that is not suppressed by rev1Δ:
The nucleotide sequences of 48 independent can1 mutations obtained from wild-type and pol3-t mutant cells revealed distinct mutation spectra (Table 3, supplemental Table 1). While the distributions of mutations among transitions, transversions, and deletions/insertions were not significantly different (P = 0.16), the fraction of deletions >3 bp in length was much greater in the pol3-t mutants (29/32) than in the wild type (2/17; P < 0.0001). Further, 25 of the 29 deletions in the pol3-t mutants were flanked by three to eight nucleotide repeats, whereas only one of the wild-type deletions shared this feature. These results are consistent with previous results demonstrating that pol3-t stimulates deletions between repetitive sequences (Gordenin et al. 1992; Tran et al. 1995, 1996; Kokoska et al. 1998, 2000; Galli et al. 2003).
While the rev1Δ allele had no significant effect on the CAN1 mutation rate (Table 2), it had a significant effect on the distribution of mutations among transitions, transversions, and deletions/insertions (Table 3; P = 0.03). However, rev1Δ did not significantly affect the fraction of deletions that were >3 bp in length (1/20; P = 0.20). Interestingly, while rev1Δ nearly completely suppressed the elevated CAN1 mutation rate conferred by pol3-t (Table 2), it restored neither the distribution of mutations among transitions, transversions, and deletions/insertions (P = 0.008) nor the elevated fraction of deletions >3 bp (30/37; P < 0.0001) to wild type. Additionally, the ratio of long deletions from the pol3-t rev1Δ double mutants that were bounded by 3- to 8-bp repeats (29/30) was not significantly different from that observed for the pol3-t single mutant (P = 0.92). These results suggest that rev1Δ may suppress the incidence of mutation in pol3-t mutant cells, but has little effect on the mechanism.
Increased rates of direct repeat recombination in pol3-t mutant strains are suppressed by rev1Δ:
Frequent replicative polymerase failure brought about by decreased processivity might be expected to increase mitotic recombination by promoting the strand invasion of the sister chromatid and repair synthesis (Navarro et al. 2007). Alternatively, polymerase failure may increase recombination by leading to the generation of DSBs through endonuclease processing at daughter strand nicks or gaps (Tishkoff et al. 1997) or upon collision between a daughter strand nick or gap and a replication fork in the next round of DNA synthesis (Navarro et al. 2007). Processes dependent on the presence of homologous sequences on the sister chromatid must occur subsequent to their generation in the S phase of the cell cycle, while other processes may also occur in G1, utilizing homologous sequences on the same chromatid.
We examined the rates of unequal sister chromatid recombination (USCR) (Fasullo and Davis 1987) in wild-type and polymerase mutant strains to determine the impact of altered polymerase activity on mitotic recombination events that are restricted to the S and G2 phases in haploid strains. Both the rev1Δ (P = 0.017) and rev7Δ (P = 0.005) alleles had significant effects on the rate of USCR, reducing it by four- and sevenfold, respectively (Table 4). This suggests that Rev1 and Pol ζ are required to propagate normal levels of USCR. Interestingly, despite having no significant effect (P = 0.065) on its own, the pol3-t allele increased the rate of USCR to wild-type levels when combined with rev1Δ (P = 0.084) and rev7Δ (P = 0.69). This suggests that the reduced levels of USCR observed in the absence of Rev1 or Pol ζ are observed only when normal Pol δ is present. The rad30Δ allele had no significant effect on USCR, either alone (P = 0.25) or in combination with pol3-t (P = 0.42), suggesting that, unlike Rev1 and Pol ζ, Pol η does not play a role in spontaneous USCR.
Deletions by recombination between nontandem direct repeats are thought to occur by a variety of mechanisms, including USCR, intrachromatid crossing over, single-ended invasion, and single-strand annealing (Schiestl and Prakash 1988; Lin et al. 1990; Belmaaza and Chartrand 1994). Unlike USCR, the other mechanisms do not require that DNA replication has proceeded through the recombination substrate, suggesting that direct repeat recombination (DRR) may not be restricted to S and G2 phases. The results of our experiments were similar to those from a number of studies that have demonstrated that the pol3-t mutation can increase the rate of DRR (Tran et al. 1997; Lobachev et al. 1998, 2000; Kokoska et al. 2000; Galli et al. 2003) as the rate was increased approximately ninefold (Table 4). The rad30Δ mutation also stimulated DRR by about fourfold, suggesting that Pol η may suppress DRR. While the rev1Δ (P = 0.28) and rev7Δ (P = 0.07) mutations alone had no significant effect on the rate of DRR, the rev1Δ mutation suppressed the hyperrecombinagenic effect of pol3-t nearly twofold (P = 0.032), while the rev7Δ mutation stimulated it nearly twofold (P = 0.001). These results suggest that Rev1 is required to observe the full stimulatory effect of pol3-t on DRR, while the presence of Pol ζ inhibits it.
The rev1Δ mutation suppresses the stimulatory effects of the pol3-t mutation on chromosome loss and interhomolog recombination:
Defects in the DNA replication apparatus have been shown previously to increase both chromosome loss (CL) and interhomolog recombination (IHR), presumably in response to an accumulation of lesions such as daughter strand nicks and gaps and DSBs (Haber 1999; Daigaku et al. 2006; Navarro et al. 2007). Increases in spontaneous IHR have previously been observed in pol3-t mutant diploids, consistent with an increase in recombinational responses to replicative polymerase failure (Galli et al. 2003). Similarly, we observed an ∼20-fold increase in the rate of IHR in pol3-t/pol3-t homozygous diploid cells (Table 5). This correlated closely with a nearly 13-fold increase in CL in the same strains, consistent with the pol3-t mutation causing an increase in recombinagenic lesions that can also disrupt chromosomal transmission, such as DSBs.
The rad30Δ allele had no significant effect on the rates of CL or IHR, either alone (P = 0.66 or 0.07) or in combination with pol3-t (P = 0.52 or 0.92), suggesting that Pol η may play no significant role in such events in diploid cells (Table 5). Interestingly, the rev7Δ mutation alone led to nearly equivalent, three- to fourfold increases in CL and IHR, suggesting that the absence of Pol ζ may increase the level of recombinagenic lesions in diploid cells. However, combining the rev7Δ mutation with the pol3-t allele did not have equivalent effects on CL and IHR. The rate of CL in the pol3-t rev7Δ double mutant was not significantly different from those in the rev7Δ single mutants (P = 0.08), indicating that rev7Δ was epistatic to pol3-t, while the rates of IHR in the pol3-t rev7Δ double mutants were not significantly different from those in the pol3-t single mutants (P = 0.58), indicating that pol3-t was epistatic to rev7Δ. This suggests that Pol ζ is required to fully stimulate CL in pol3-t mutant cells, but does not significantly affect the impact of pol3-t on IHR.
In contrast to rev7Δ, the rev1Δ allele itself did not have equivalent effects on CL and IHR (Table 5), increasing CL by nearly threefold, but having no significant effect on IHR (P = 0.07). However, rev1Δ did have nearly equivalent, suppressive effects on the stimulation of CL and IHR by pol3-t, as the rates of both events were reduced between three- and fourfold in pol3-t rev1Δ double-mutant diploids relative to those in pol3-t single-mutant diploids. Therefore, unlike Pol ζ, Rev1 is required to fully stimulate both CL and IHR in response to the defective Pol δ encoded by pol3-t.
The connection between dysfunctional DNA replication and genome instability has been firmly established in yeast by the elevated rates of many types of mutagenic and clastogenic events observed in a variety of DNA replication mutants (Aguilera and Klein 1988; Gordenin et al. 1992; Ruskin and Fink 1993; Reagan et al. 1995; Ohya et al. 2002; Meyer and Bailis 2007). In particular, mutations that disrupt the polymerase function of Pol δ confer increased rates of mutation and recombination (Gordenin et al. 1992; Tran et al. 1995, 1996, 1997; Kokoska et al. 1998, 2000; Lobachev et al. 1998, 2000; Galli et al. 2003), consistent with incompletely or improperly replicated DNA serving as a source for spontaneous mutation and genome rearrangement. Importantly, Pol δ has also been implicated in post-replication repair (Giot et al. 1997; Torres-Ramos et al. 1997; Galli et al. 2003), suggesting that it could be involved in facilitating mutagenic and recombinagenic responses to lesions created by replicative polymerase failure. The dual role of the DNA replication apparatus in replication and repair suggests that altered levels of mutation and recombination observed in DNA replication mutants may reflect defects in DNA replication, DNA repair, or both.
The pol3-t mutation confers a temperature-sensitive growth defect and increased rates of intrachromosomal deletion consistent with a decrease in the processivity of Pol δ (Tran et al. 1995; Lobachev et al. 1998, 2000; Kokoska et al. 2000). Decreased processivity might be expected to lead to an increase in polymerase pausing during replication, which could yield an increase in daughter strand nicks and gaps that have been associated with increased genome instability (Cox 1999; Lehmann and Fuchs 2006; Nagaraju and Scully 2007). This is consistent with the increased rates of mutation, recombination, and chromosome loss observed in this (Tables 2, 4, and 5) and previous studies (Gordenin et al. 1992; Tran et al. 1995, 1996, 1997; Kokoska et al. 1998, 2000; Lobachev et al. 1998, 2000; Galli et al. 2003). However, Pol δ has also recently been implicated in the repair of DSBs by homologous recombination (Lydeard et al. 2007; Maloisel et al. 2008), and certain homologous recombination mutants display increased rates of spontaneous mutation, recombination, and chromosome loss (Mortimer et al. 1981; Klein 2001; Yoshida et al. 2003; Navarro et al. 2007). Therefore, the increased genome instability observed in pol3-t mutant strains instead may result from increased steady-state levels of DSBs and other lesions that accumulate due to a defect in their repair.
Assaying mutation at the CAN1 locus is advantageous because it permits the quantitation and characterization of a variety of mutation types (Whelan et al. 1979; Tishkoff et al. 1997). We observed that the mutation frequency at CAN1 in the pol3-t mutant was elevated ∼10-fold (Table 2), consistent with previously published results (Galli et al. 2003; Northam et al. 2006). The increase in mutation rate was accompanied by a striking increase in the frequency of deletions >3 bp that were bordered by repetitive sequences of 3–8 bp. This is highly reminiscent of the results of previous experiments documenting increases in reversion events involving deletions between short repeats (Tran et al. 1995, 1996; Kokoska et al. 1998) and is consistent with the view that pol3-t promotes increased Pol δ failure and promiscuous reassociation with the template during replication or repair synthesis. Further, it establishes that, like mutants defective for the gene encoding the lagging strand maturation and DNA repair factor Rad27 (Tishkoff et al. 1997), pol3-t mutants display a signature mutation.
The results of our recombination assays (Tables 4 and 5) generally reflect those of previous experiments that documented substantial increases in DRR and IHR in the pol3-t mutant (Gordenin et al. 1992; Kokoska et al. 2000; Galli et al. 2003). This is consistent with pol3-t conferring an increase in daughter strand nicks and gaps and other recombinagenic lesions. Interestingly, we observed no significant change in the rate of USCR in the pol3-t mutant (Table 4). While nuclease-catalyzed DSBs can stimulate USCR (Fasullo et al. 1998), spontaneous USCR has been proposed to occur by strand invasion and repair synthesis from the sister chromatid subsequent to replicative polymerase pausing (Navarro et al. 2007). Gangavarapu et al. (2007) have suggested that RAD52-dependent post-replication repair following disruption of lagging strand synthesis may occur by a similar mechanism. Although Pol δ has been implicated in lagging strand synthesis (Garg and Burgers 2005b; Nick McElhinny et al. 2008), the pol3-t mutant is unlike rad27Δ mutants that display a 46-fold increase in the rate of USCR (Navarro et al. 2007) along with increased levels of daughter strand nicks and gaps (Vallen and Cross 1995; Parenteau and Wellinger 1999). This suggests that the pol3-t mutation may result in the accumulation of fewer daughter strand nicks and gaps than rad27Δ or that it inhibits the utilization of these lesions for USCR. The wild-type rates of USCR observed when pol3-t was combined with the rev1Δ and rev7Δ alleles (Table 4) suggests that pol3-t may channel these lesions away from Rev1- and Pol ζ-dependent USCR into other repair pathways.
Translesion polymerases have been proposed to function in circumstances where the replicative polymerases cannot function because of rigid constraints on their ability to use altered or damaged DNA templates (Prakash et al. 2005). The translesion polymerases have been proposed to replace the replicative polymerase at the stalled replication fork in a carefully orchestrated process (Friedberg et al. 2005), allowing the bypass of lesions in a manner that frequently introduces mutations (Kunkel 2003). However, it is unclear what role these auxiliary polymerases may play in pol3-t mutants, whose defective Pol δ may affect primarily lagging strand synthesis (Garg and Burgers 2005b; Nick McElhinny et al. 2008), the interruption of which could generate daughter strand nicks or gaps, but may not result directly in replication fork stalling.
Pol η was been tentatively implicated in spontaneous mutagenesis in yeast (McDonald et al. 1997; Roush et al. 1998), as has Pol ζ (Quah et al. 1980; Roche et al. 1994; Kunz et al. 1998; Endo et al. 2007). However, the extent to which this is a response to wild-type levels of spontaneous DNA lesions and/or spontaneous polymerase failure is unclear. Pol η, Pol ζ, and Rev1 have not been reported to be required for spontaneous recombination in yeast, which is in contrast to studies with higher eukaryotic systems (Gan et al. 2008). Our data are consistent with Pol η, Pol ζ, and Rev1 playing minor roles in spontaneous mutation, recombination, and chromosome loss (Tables 2, 4, and 5), as rates of these events were, for the most part, similar to wild type in the rev1Δ, rev3Δ, rev7Δ, and rad30Δ single-mutant strains. The most interesting observation was the four- and sevenfold decreased levels of USCR displayed by the rev1Δ and rev7Δ mutants, consistent with Rev1 and Pol ζ facilitating spontaneous recombination events between sisters. This may be the first evidence of a requirement for translesion polymerases in the propagation of spontaneous recombination in yeast. Given that USCR can occur only subsequent to completion of replication through the USCR substrate, this result strongly suggests that Rev1 and Pol ζ participate in spontaneous recombination in the S or G2 phases of the cell cycle, consistent with data concerning the abundance and/or likely function of these proteins in the G2 phase (Waters and Walker 2006). These results are also consistent with data supporting the function of Rev1 and Pol ζ during sister-chromatid exchange in vertebrate cells (Okada et al. 2005).
The mutation and recombination data reported here indicate that null mutations in the translesion polymerase genes RAD30, REV3, and REV7 do not suppress the genome-destabilizing effects of pol3-t (Tables 2, 4, and 5). This contradicts the results of previous studies that demonstrated that a rev3Δ mutation is able to suppress the elevated mutation rates conferred by certain pol3 alleles, including pol3-t (Pavlov et al. 2001; Northam et al. 2006). These contradictory results are likely to reflect differences in the genetic backgrounds of the strains used in the different studies. However, the current results suggest that pol3-t is, at least in certain contexts, capable of exerting its mutagenic and recombinagenic effects independently of the action of Pol η and Pol ζ.
In contrast to Pol η and Pol ζ, the epistasis interactions between rev1Δ and pol3-t indicate that Rev1 is required to fully observe the genome-destabilizing effects of the pol3-t allele, as the increased rates of mutation, recombination, and chromosome loss observed in the pol3-t single mutants are all significantly suppressed in the pol3-t rev1Δ double mutants (Tables 2, 4, and 5). This requirement is most clearly demonstrated by the CAN1 mutation rate, where the substantial stimulatory effect of pol3-t is nearly completely suppressed by rev1Δ in the pol3-t rev1Δ double mutant. Significantly, Rev1 is seen acting independently of Pol ζ in promoting spontaneous mutagenesis, where previously they had been known to function together (Lawrence and Maher 2001; Lawrence 2002) and with Rev1 playing a structural role (Acharya et al. 2005; Prakash et al. 2005). Interestingly, although translesion replication is believed to be native to all eukaryotes, Caenorhabditis elegans lacks a REV3 gene but possesses a REV1 gene, suggesting that Rev1 may collaborate with another polymerase, such as Pol δ, to propagate translesion synthesis (Lawrence and Maher 2001). It is clear from studies of the mouse that Rev1 has the capacity to interact with multiple polymerases (Guo et al. 2003). In keeping with this scenario, it is tempting to speculate that Rev1 may facilitate mutagenesis in pol3-t mutant yeast strains by aiding the association of the defective Pol δ with daughter strand nicks and gaps during repair synthesis. However, the very similar can1 mutation spectra displayed by the pol3-t single and pol3-t rev1Δ double mutants (Table 3, supplemental Table 1) suggest that the absence of Rev1 may reduce but not eliminate the association of Pol δ with these lesions.
We suggest that increased mutation, recombination, and chromosome loss may have a common mechanistic origin in pol3-t mutant cells, where each is the result of Rev1 helping to engage a dysfunctional Pol δ for the repair of daughter strand nicks or gaps during G2. Mutations may result when Pol δ that is prone to promiscuous reassociation with the template is engaged to repair the lesions. Recombination and chromosome loss may result when Pol δ that is prone to premature failure is engaged and falls off before completing repair, leading to the persistence of nicks and gaps that are transformed into recombinagenic DSBs during the subsequent S phase (Navarro et al. 2007).
This study suggests that the genome instability that results from elevated levels of spontaneous replicative DNA polymerase failure is the result of the tandem action of primary replicative and auxiliary translesion polymerases. These observations may have clinical significance as high levels of spontaneous replicative polymerase failure may also occur as the result of the administration of chemotherapeutic drugs such as hydroxyurea, which drive down levels of dNTPs by inhibiting ribonucleotide reductase activity (Elford 1968; Cory et al. 1980). Perhaps the elevated levels of secondary cancers observed in patients that have received such drugs are due to an increase in genome instability from Rev1-mediated responses to replicative polymerase failure. In such a scenario, pharmacological disruption of Rev1 activity during the administration of chemotherapeutic drugs may reduce genome instability and the incidence of secondary cancers.
We thank J. McDonald, R. Woodgate, and D. Gordenin for strains and plasmids. We also thank D. Tamae for assistance and J. Termini, L. D. Finger, members of the Bailis laboratory, and several anonymous reviewers for stimulating discussions and comments. This work was supported by a U. S. Public Health Service grant (GM057484 to A.M.B.), a Rose Hills Foundation fellowship (J.V.M.), a Metcalfe Foundation fellowship (M.C.S.), funds from the Department of Defense/Department of the Interior (1435-04-06-GT-63257 to G.M.M.), as well as funds from the Beckman Research Institute of the City of Hope and the City of Hope National Medical Center. This manuscript was submitted for publication with the understanding that the United States government is authorized to reproduce and distribute reprints for government purposes. The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the U. S. government.
↵1 Present address: Anderson School at UCLA, Global Economics and Management Area, 110 Westwood Plaza; C513 Entrepreneurs Hall, Los Angeles, CA 90095-1481.
Communicating editor: S. Keeney
- Received April 1, 2008.
- Accepted May 18, 2008.
- Copyright © 2008 by the Genetics Society of America