In Vivo Consequences of Putative Active Site Mutations in Yeast DNA Polymerases , , , and

Corresponding author: Youri I. Pavlov, Laboratory of Molecular Genetics, Bldg. 101, Rm. 332, National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709., pavlov{at}niehs.nih.gov (E-mail)

Communicating editor. P. L. FOSTER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Several amino acids in the active site of family A DNA polymerases contribute to accurate DNA synthesis. For two of these residues, family B DNA polymerases have conserved tyrosine residues in regions II and III that are suggested to have similar functions. Here we replaced each tyrosine with alanine in the catalytic subunits of yeast DNA polymerases , , , and and examined the consequences in vivo. Strains with the tyrosine substitution in the conserved SL/MYPS/N motif in region II in Pol or Pol are inviable. Strains with same substitution in Rev3, the catalytic subunit of Pol, are nearly UV immutable, suggesting severe loss of function. A strain with this substitution in Pol (pol1-Y869A) is viable, but it exhibits slow growth, sensitivity to hydroxyurea, and a spontaneous mutator phenotype for frameshifts and base substitutions. The pol1-Y869A/pol1-Y869A diploid exhibits aberrant growth. Thus, this tyrosine is critical for the function of all four eukaryotic family B DNA polymerases. Strains with a tyrosine substitution in the conserved NS/VxYG motif in region III in Pol, -, or - are viable and a strain with the homologous substitution in Rev3 is UV mutable. The Pol mutant has no obvious phenotype. The Pol (pol2-Y831A) mutant is slightly sensitive to hydroxyurea and is a semidominant mutator for spontaneous base substitutions and frameshifts. The Pol mutant (pol3-Y708A) grows slowly, is sensitive to hydroxyurea and methyl methanesulfonate, and is a strong base substitution and frameshift mutator. The pol3-Y708A/pol3-Y708A diploid grows slowly and aberrantly. Mutation rates in the Pol, -, and - mutant strains are increased in a locus-specific manner by inactivation of PMS1-dependent DNA mismatch repair, suggesting that the mutator effects are due to reduced fidelity of chromosomal DNA replication. This could result directly from relaxed base selectivity of the mutant polymerases due to the amino acid changes in the polymerase active site. In addition, the alanine substitutions may impair catalytic function to allow a different polymerase to compete at the replication fork. This is supported by the observation that the pol3-Y708A mutation is recessive and its mutator effect is partially suppressed by disruption of the REV3 gene.

THE fidelity of chromosomal DNA replication depends on the base selectivity of DNA polymerases, on exonucleolytic proofreading, and on DNA mismatch repair. Genetic studies of proofreading- and mismatch repair-deficient strains of yeast and Escherichia coli (e.g., MORRISON and SUGINO 1992 ; MORRISON et al. 1993 ; SCHAAPER 1993 ) indicate that, among these steps, the greatest contribution to a low spontaneous mutation rate is provided by the high base selectivity of replicative DNA polymerases. The major replicative DNA polymerases catalyze DNA synthesis with error rates ranging from 10^-4 to 10^-7 (reviewed in ROBERTS and KUNKEL 1996 ; KUNKEL and BEBENEK 2000 ). At least three DNA polymerases are required for replication of the eukaryotic nuclear genome, DNA polymerase (Pol), Pol, and Pol (SUGINO 1995 ; BURGERS 1998 ; WAGA and STILLMAN 1998 ). Recent studies (reviewed in FRIEDBERG and GERLACH 1999 ; FRIEDBERG et al. 2000 ; GOODMAN and TIPPIN 2000 ) suggest that additional, specialized DNA polymerases also operate under special circumstances. For example, Pol and Pol are required when the replication machinery encounters lesions in DNA (LAWRENCE 1996 ; JOHNSON et al. 1999 ; MASUTANI et al. 1999 ) and Pol4 may be involved in nonhomologous end joining (WILSON and LIEBER 1999 ).

Our understanding of the mechanisms responsible for the high base selectivity of polymerization has been greatly facilitated by structure-function studies of DNA polymerases. The crystal structures of family A (Pol I family) DNA polymerases (E. coli Pol I, Taq Pol, T7 Pol) suggest that the binding pocket for the newly forming base pair snugly accommodates correct Watson-Crick base pairs but excludes mispairs with abnormal geometry (reviewed in KUNKEL and BEBENEK 2000 ). The importance of this binding pocket to base selection is indicated by functional studies of mutant DNA polymerases whose fidelity is increased or decreased by amino acid substitutions in or near the DNA polymerase active site (MINNICK et al. 1999 ). The present study was initiated on the basis of observations with two mutants of the large Klenow fragment of E. coli DNA polymerase I. These mutants contain alanine substituted for residues that are invariant in family A polymerases, Glu710 and Tyr766. Both mutant enzymes have reduced DNA synthesis fidelity in vitro (CARROLL et al. 1991 ; BELL et al. 1997 ; MINNICK et al. 1999 ). This is consistent with the structural location of these two amino acids at the polymerase active site and their importance in geometric selection of correct base pairs.

Just as for family A enzymes, the polymerase domains of three family B polymerases (Pol family), RB69 Pol (WANG et al. 1997 ) and two Archea polymerases (HOPFNER et al. 1999 ; ZHAO et al. 1999 ), are composed of fingers, thumb, and palm subdomains. The palm subdomains of these enzymes contain catalytic carboxylates in positions that are thought to be structurally equivalent to those in Pol I family polymerases. When these catalytic residues in Klentaq Pol (a family A homolog of Klenow fragment Pol) and RB69 Pol are superimposed, RB69 residues Tyr416 and Tyr567 are found in positions that are similar to those of the Glu710 and Tyr766 residues of Klenow fragment Pol mentioned above. This led to the suggestion (WANG et al. 1997 ) that Tyr416 and Tyr567 in RB69 Pol may have functions that are similar to those of Glu710 and Tyr766 in Klenow fragment Pol. Interestingly, these two tyrosine residues are conserved in family B polymerases from bacteriophage to man, including the yeast replicative Pol, -, and - and the specialized Pol (Fig 1). This information led to this study, whose objective was to extend our previous structure-function analyses of DNA polymerase fidelity in vitro to an in vivo setting in eukaryotic cells. Here we examine the importance of two conserved tyrosine residues in the three yeast replicative DNA polymerases and in the Pol involved in UV mutagenesis. These represent all the known yeast family B DNA polymerases. To obtain a more global view of mutator effects conferred by these mutant enzymes, we first constructed a novel yeast strain to concomitantly measure rates for various types of mutations at five different loci. We then introduced into this strain polymerase alleles encoding alanine instead of either of the two tyrosines in the catalytic subunits of Pol, -, -, and -. We examined the effects of these mutations on growth, sensitivity to DNA damaging agents, and spontaneous (for replicative polymerases) or UV mutagenesis (for Pol). These effects were examined in an otherwise wild-type background and in combination with defects in DNA mismatch repair or specialized DNA polymerases.

View larger version (37K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Alignment of amino acid sequences of regions II (A) and III (B) of yeast DNA polymerases with RB69 and T4 polymerases, polymerase I, and Taq polymerase. Alignment is based on the work of DELARUE et al. 1990 . Region II has a different aligning of E710 of Klenow and Taq polymerases as suggested by crystal structures of RB69 and Klenow DNA polymerase superimposition (WANG et al. 1997 ). Strictly conserved amino acids are boxed. Amino acids on which this study is focused are underlined. RB69GP43, DNA polymerase of bacteriophage RB69; T4GP43, DNA polymerase of bacteriophage T4; SCPOL1, SCPOL2, SCPOL3, and SCREV3, corresponding yeast Saccharomyces cerevisiae DNA polymerases Pol, Pol, Pol, and Rev3; ECPOLI, E. coli DNA polymerase I; TAQPOLI, Thermus aquaticus DNA polymerase I.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plasmids:
Chromosomal DNA polymerase mutations were generated by an integration-excision method (see MORRISON et al. 1991 ). Plasmid p173 (Fig 2; KIRCHNER et al. 2000 ) contains the BamHI-BspEI fragment of POL2 cloned in the BamHI-AvaI site of pFL34* [similar to pFL34 but with URA3 in the other orientation (TRAN et al. 1995 , TRAN et al. 1997 )]. Plasmid p174 (kindly provided by Dr. H. Tran) contains the BamHI-SstI fragment of POL2 cloned in the BamHI-SacI site of pFL34*. These plasmid variants with desired DNA sequence changes were cut by AgeI (p173), Acc65-I, or Tth111-I (p174) prior to transformation, to create the POL2 (Pol) gene mutations. For creation of mutations in the POL3 (Pol) gene, plasmid p170 (KOKOSKA et al. 1998 ) variants with desired mutations made as described below were cut by HpaI or BseRI prior to transformation. For delivery of mutations into the POL1 (Pol) gene, a new plasmid was constructed. First, the BglII fragment with the URA3 gene from plasmid p170 was made blunt ended using Klenow fragment of DNA polymerase I and then cloned into the BstZ17-I site of the low copy number plasmid pMBL18 (also with blunt ends; NAKANO et al. 1995 ), thus generating plasmid pYIYI18. In this construct, the -complementing lacZ region with the polylinker remains intact. Transformants with a Ura⁺ phenotype were selected by complementing the pyrF mutation in E.coli strain MC1066 (CASADABAN et al. 1983 ). Active site mutations were created as described below in pPol1-8 (BUDD and CAMPBELL 1987 ). Then, 3.7-kb HindIII fragments with the N-terminal part of the POL1 gene from pPol1-8 mutant variants were cloned into the HindIII site of pYIYI18 to generate pYIAL31-pol1-Y869A and pYIAL31-pol1-Y951A. These plasmids were cut with either Eco47-III or BspEI to target integration into the POL1 locus. For delivery of mutations into the REV3 (Pol) gene, the plasmid pRevLCav2 was constructed. A 2.3-kb AvaII fragment containing the C-terminal part of the REV3 from the plasmid pJA6 (MORRISON et al. 1989 ) was cloned into SmaI site of pYIYI18. Site-directed mutagenesis was performed as described below and pRev3Cav2 variants with desired mutations were cut by SnaBI prior to transformation to target integration into the REV3 gene. Plasmid pAM58 (MORRISON et al. 1993 ) was used to disrupt the PMS1 gene and plasmid pAM56 (A. MORRISON, personal communication) was used to disrupt REV3 with the LEU2 gene.

View larger version (25K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Fragments of DNA polymerase genes used for polymerase mutant construction: (A) POL1, (B) POL2, (C) POL3, and (D) REV3. A schematic map of DNA polymerase genes is shown. Open reading frame sequences are represented by rectangles. The thick lines below the sequence depict DNA polymerase gene fragments cloned in integrative plasmids. Names of the corresponding plasmids are below these lines. Restriction sites (nucleotide numbering starts from the first ATG codon of the open reading frame) are shown above the polymerase regions. Sites of amino acids that were targeted for site-directed mutagenesis are shown as vertical lines crossing open reading frames (ORFs). Numbers above these bars refer to a third nucleotide in the codon coding for the corresponding amino acids. Numbers are shown in boldface below the ORF box.

Yeast strains:
Yeast strains used in this study are listed in Table 1. To simultaneously determine forward mutation rates as well as rates of reversion of a frameshift mutation and rates of transitions and transversions, we constructed a new yeast strain, 8C-YUNI101. Standard methods of yeast genetics were used (ROSE et al. 1990 ). We began with strain CG379-3-29RL, a derivative of the strain CG379 (SHCHERBAKOVA et al. 1996 ) with the ura3-29 allele inserted into chromosome III in a RL orientation (MORRISON and SUGINO 1994 ) in the BIK1 locus (see MORRISON and SUGINO 1994 ; SHCHERBAKOVA and PAVLOV 1996 and Saccharomyces Genome Database). A pol3-01 variant of this strain (exoD-CG379-3-29RL) was crossed to S111-2-11 (BUDD and CAMPBELL 1993 ) to generate the diploid strain SPJ1. A meiotic segregate of this diploid, 5B-SPJ1, was crossed to h1-27B-SPJ5, which is a His⁺ revertant (generated by transformation with a wild-type HIS3 PCR fragment) of 27B-SPJ5. 27B-SPJ5 resulted from a cross of CG379 to W303-1A (THOMAS and ROTHSTEIN 1989 ). The cross between 5B-SPJ1 and 27B-SPJ5 generated the diploid strain YUNI101. Tetrad analysis of a YUNI101 diploid gave 8C-YUNI101 (MATa his7-2 leu2-3,112 ura3 bik1::ura3-29RL trp1-1_UAG ade2-1_UAA). PCR analysis showed that this segregate does not have a mutation in RAD5 gene that could be inherited from W303-1A (see MCDONALD et al. 1997 ).

Strain 8C-YUNI101 permits measurement of the forward mutation rate at the CAN1 locus, the rate of reversion of the ura3-29 missense mutation [originally induced by the base analogue HAP, reversion occurs mainly by intragenic events (SHCHERBAKOVA and PAVLOV 1996 and our results below)], reversion of nonsense mutations trp1-1 and ade2-1 that occurs via both intra- and intergenic events, and reversion of the his7-2 mutant allele that occurs mainly via +1 or -2 frameshifts in a homopolymeric run (SHCHERBAKOVA and KUNKEL 1999 and our results below). Other strains used included YUNI190, a diploid strain resulting from crossing 8C-YUNI101 and CG379-3-29RL; YUNI192, a diploid resulting from crossing Del708-8C-YUNI101 and Del708-CG379-3-29RL (both containing the pol3-Y708A mutation); and YUNI200 diploid series resulting from crossing 8C-YUNI101 and E134 or their polymerase and mismatch repair defective derivatives.

Mutants of the basic strains were prepared using the integration-excision method and by targeted gene disruption via transformation with fragments of plasmids or PCR fragments carrying a selectable kanMX cassette or a hygromycin B-resistance marker flanked by short sequence homology to the target gene (WACH et al. 1994 ; GOLDSTEIN and MCCUSKER 1999 ). The REV3 gene was disrupted either by LEU2 using the pAM56 plasmid cut by XbaI or by PCR fragments with antibiotic resistance markers. The RAD30 and POL4 genes were disrupted by the kanMX cassette. The primers used are listed in Table 2A. Double and triple mutants were constructed by sequential transformations with corresponding plasmids. The last mutation generated was always in the DNA mismatch repair genes, since these mutations cause a strong mutator phenotype.

Generation of DNA polymerase point mutations:
Mutations were introduced into relevant plasmids (Fig 2) by PCR using the primers shown in Table 2B. Plasmid DNA from four to six individual bacterial transformants after mutagenesis was purified and sequenced. At least two independent isolates containing the correct mutant sequence and no other changes in 0.5–1 kb in the vicinity of the alanine codon were used for constructing the yeast mutant strains. Consistent results were obtained with duplicate isolates in every case. The polymerase mutations were introduced into the chromosome by integration of appropriately cut plasmids with selectable URA3 marker (described above) to target integration into polymerase genes (see MORRISON et al. 1991 ). Initial transformants possess duplication of part of a polymerase gene. Integrants were grown on 5'-fluoroorotic acid containing medium to select for excision of URA3 and reconstitution of the original chromosomal structure. The presence of the mutation in the chromosomal DNA was detected by PCR using primers with 3'-ends complementary to either the mutant or wild-type sequence. We used a set of four primers for each DNA polymerase to detect both E710 and Y766 homolog mutations (Table 2, C–F, set 1). The strategy is explained here for Pol. Primer al 869Y WTL (Table 2C) is complementary to the wild-type sequence at the pol1-Y869A site. Primer al 869A L is complementary to the mutated sequence at the pol1-Y869A site. Primer al 951Y WTR is a reverse complement to the wild-type sequence at the pol1-Y951A mutation site and primer al 951A R is a reverse complement to mutated sequence at the pol1-Y951A mutation site (3'-end bases critical for discrimination are shown in boldface in Table 2). A combination al 869Y WTL/al951Y WTR amplifies a DNA fragment with the wild-type sequence of POLI; the al 869A L/al 951Y WTR combination amplifies a DNA fragment with the pol1-Y869A mutation, and the al 869Y WTL/al 951A R combination amplifies the pol1-Y951A mutation. The wild-type strain gives a positive signal only with the first pair of primers, the majority of initial integrants give signals with the wild-type and mutant primers, and the final mutants give a positive signal only with primers complementary to the mutant sequence. The same approach was taken with the other mutations, using the primers listed in Table 2D, Table 2E, and Table 2F. Positive clones were finally analyzed by DNA sequencing of the relevant portion of the polymerase gene. Diploid transformants with the pol3-Y613A (Pol) and pol2-Y645A (Pol) mutations in the heterozygous state were constructed in a similar manner. However, since selection for loss of the URA3 gene due to plasmid excision is obscured by mitotic recombination, the initial Ura⁺ transformants were examined by PCR with a different set of primers to confirm that they have a mutation in a full-length copy of the polymerase gene (Table 2D and Table 2E, set 2). Primers ZABAM and ZAECOV are complementary to portions of polymerases and genes, respectively, that are beyond the fragment cloned in the plasmids used to create mutations (Fig 2). In combination with primers complementary to mutated or wild-type sequence these primers permit detection of the mutated sequence in the full-length portion of a polymerase gene. Such diploid transformants were plated onto sporulation medium and tetrads were dissected. Resulting colonies on YPD were examined after 4 days of incubation at 30°.

Qualitative tests for sensitivity to UV irradiation and chemical treatments:
Serial 10-fold dilutions of each strain (starting from 5 x 10⁷ cells/ml) were prepared in 96-well microtiter plates and plated with a 48-prong replicator (Sigma, St. Louis) onto YPD medium or YPD containing hydroxyurea (HU) or methyl methanesulfonate (MMS). For the UV irradiation sensitivity test, cells on plates were UV irradiated using two germicidal lamps at 1.5 J/m²/sec at the dose indicated in the legend to Fig 3B.

View larger version (131K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Growth of yeast strains with polymerase mutations. (A) Tetrad analysis of control diploid, YUNI200 (top), and diploids heterozygous for pol2-Y645A (YUNI200-hetEP, bottom left) and for pol3-Y613A (YUNI200-hetDEL, bottom right). (B) Sensitivity of yeast polymerase mutants to HU, UV light, and MMS. Serial 10-fold dilutions of each strain starting from 5 x 107 cells/ml were prepared in a 96-well microtiter plate and plated onto corresponding medium with a 48-prong replicator (Sigma). Each column on the plate represents a dilution of one strain: wild type, pol1-Y951A, pol1-Y831A, pol2-Y831A, and pol3-Y708A.

Quantitative tests for mutability by UV irradiation:
Appropriately diluted cells were plated onto YPD or selective medium and UV irradiated at doses indicated in the legend to Fig 6. Mutant frequencies were then calculated as the ratio of the number of mutants on selective plates to number of colonies on YPD medium multiplied by the dilution factor.

View larger version (63K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Growth and mating phenotypes of diploids homozygous for DNA polymerase mutations. (A) Genetic crosses of strains with DNA polymerase mutations. Plate shown is replica plating of the cross from YPD to medium selective for hybrids. Diploids could be seen as rectangle areas of growth on the intersections of slightly visible traces of nongrowing haploid parent strains. (B) Mating of wild-type (left) and pol3-Y708A homozygous (right) diploids with mating-type testers. Individual clones of wild-type diploid (YUNI190) and clones of diploid arising from cross of two haploid pol3-Y708A mutants (YUNI192, bottom right diploid in Fig 4A) were crossed on YPD to standard mating-type tester strains and replica plated onto medium selective for hybrids.

View larger version (20K): [in this window] [in a new window] [Download PPT slide]   Figure 5. Effects of mutations in specialized DNA polymerases on mutator effect of the pol3-Y708A in a mismatch repair-defective pms1 strain. (A) Rates of Ura+ reversion. (B) Rates of His+ reversion. Rates are medians for 9–27 independent cultures. Error bars represent 95% confidence intervals.

View larger version (17K): [in this window] [in a new window] [Download PPT slide]   Figure 6. UV irradiation effect on survival and mutagenesis in strains with mutations in the REV3 gene. (A) Survival (percentage). (B) Frequencies of Canr mutants. (C) Frequencies of Ade+ mutants. Mean values from at least three experiments. Error bars are standard errors. —, REV3, - - -, rev3-Y980A; ···, rev3::LEU2; -·-·-, rev3-Y1093A.

Measurement of spontaneous mutation rates:
The fluctuation tests to determine spontaneous mutation rates were performed using at least nine independent cultures. Single 2-day-old colonies from YPD plates were inoculated in 5 ml of liquid YPD medium and were grown with strong aeration for 2 days and processed as described (SHCHERBAKOVA and KUNKEL 1999 ).

Revertant sequencing:
Independent Ura⁺ or His⁺ revertants were grown as small patches on YPD plates, regions of corresponding genes were amplified by PCR on whole cells, amplified DNAs were purified by QIAGEN (Valencia, CA) PCR purification kit and sequenced using an automated PE ABI377 DNA sequencer.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Tester strain to monitor mutator effects:
For this study we constructed a new yeast strain, 8C-YUNI101, in which mutation rates can be measured at five different genetic loci. This includes forward mutations to canavanine resistance, reflecting a variety of substitution, frameshift, and more complex mutations (for example, see CHEN et al. 1998 ). 8C-YUNI101 harbors the his7-2 allele containing a single base deletion in a run of 8 A·T base pairs, which can revert by addition of a base pair, loss of two base pairs or, more rarely, complex events (SHCHERBAKOVA and KUNKEL 1999 and this study). 8C-YUNI101 harbors the ura3-29 allele, which reverts via three specific base pair substitutions at a single G·C base pair in a TCT codon encoding for serine (see SHCHERBAKOVA and PAVLOV 1996 and this study). Finally, 8C-YUNI101 also contains the trp1-1 and ade2-1 nonsense alleles, which revert presumably via a broad range of base pair substitution mutations. The ade2-1 mutation imparts on this strain the growth-dependent accumulation of red pigment, which makes it possible to color-select slow-growing variants, such as respiratory-defective petite mutants. The concomitant analysis of five different mutational markers in one strain was informative regarding the mutator specificity of the mutant DNA polymerase alleles examined here.

Phenotypes conferred by replicative Pol mutations in haploid yeast strains:
We used plasmids with truncated DNA polymerase , , or genes to introduce coding sequence changes that replaced either of the two conserved tyrosine residues (Fig 1) with alanine. These six different mutant alleles were then independently integrated into the genome of strain 8C-YUNI101. To facilitate identification of strains with the desired allele, we controlled the construction process at the initial integration step and during the subsequent pop out of the wild-type allele, using PCR with allele-specific primers (see MATERIALS AND METHODS). This substantially reduced the number of clones to be analyzed and permitted identification of mutant clones without knowledge of phenotypes. It also proved to be an invaluable tool for constructing polymerase mutants in the heterozygous state. The data obtained with mutants are summarized in Table 3, which can also be used as a guide for mutant allele nomenclature.

Transformation of the haploid yeast strain 8C-YUNI101 with plasmids designed to generate the Y645A Pol mutant and the Y613A Pol mutant did not yield the desired mutant strains, suggesting that these two changes are lethal in this haploid strain. To test this, we integrated the plasmids into the diploid strains YUNI190 or YUNI200 and verified the presence of the mutations in the heterozygous state in a full size chromosomal copy of the gene by PCR analysis (see MATERIALS AND METHODS). Tetrad analysis of the two mutant strains yielded 2:2 segregation for viability (Fig 3A). All viable spores were Ura^-. These results suggest that the Y645A mutation in Pol and the Y613A mutation in Pol are incompatible with vegetative growth. Note that some spores with the Y645A mutation in Pol formed tiny residual colonies (Fig 3A), suggesting that the effect of this mutation is less severe than for the Y613A Pol mutation.

In contrast to the above effects, strains harboring the Y869A or Y951A mutations in Pol, the Y708A mutation in Pol, or the Y831A mutation in Pol were all viable. The Y951A Pol strain and the Y831A Pol strain grew normally on YPD medium (Fig 3B, control plates and data not shown). In contrast, the Y869A Pol strain and the Y708A Pol strains grew more slowly and had a reduced cell number in stationary phase cultures when compared to the wild-type strain (cell titers were 2–3 x 10^-8 for wild-type and 0.9–1 x 10^-8 for Y708A Pol or Y869A Pol strains). None of the four mutant strains was highly sensitive to UV irradiation (Fig 3B). In contrast, three of the four mutant strains (pol1-Y869A, pol2-Y831A, and pol3-Y708A) were more sensitive than the wild-type strain to treatment with HU, a known inhibitor of DNA replication. The Y869A Pol mutant and Y708A Pol mutant strains were also sensitive to 0.01% MMS (Fig 3B). These initial qualitative analyses indicate that the phenotypic consequences of these six polymerase mutations are in the following order of increasing severity (for summary, see Table 3): Y951A Pol, Y831A Pol, Y869A Pol, Y708A Pol, Y645A Pol (microcolonies), and Y613A Pol (inviable).

Diploid strains heterozygous or homozygous for replicative Pol mutations:
Among the five mutant polymerase alleles that affected one or more phenotypes as haploids (the exception being pol1-Y951A, which had no phenotypic consequence), four did not yield detectable phenotypic effects on growth rate, MMS and HU sensitivity, or mutagenesis in the presence of the wild-type Pol allele in heterozygous diploid strains. Thus, these alleles are recessive (Table 3). However, as shown below, the pol2-Y831A mutation was semidominant as evidenced by elevated frameshift mutation rate in a mismatch repair-defective strain. Diploid strains homozygous for pol1-Y951A and pol2-Y831A grow normally, like the corresponding wild-type diploid strain. Homozygosity for pol3-Y708A produced very slow growth (Fig 4A, lower right diploid), and this was also seen with the pol1-Y869A mutation (data not shown). Cells in the colonies of these slow-growing diploid strains had abnormal morphology (not shown). These diploid strains are all MATa/MAT and thus would not normally mate, as is the case for the wild-type diploid strain (left plate of Fig 4B). However, almost all single-colony isolates of the homozygous mutant diploid strains were able to mate with one or both of mating-type testers (Fig 4B). This indicates that they had either lost chromosome III or were losing it frequently while the colonies were growing (MORTIMER et al. 1981 ). A possibility that these mating clones were generated by mitotic recombination is unlikely since mitotic recombination is not affected by the pol3-Y708A mutation to a level that could explain the observation (Y. I. PAVLOV, unpublished data). The pol1-Y869A and pol3-Y708A mutants had a reduced ability to maintain centromeric plasmids (data not shown). Taken together, these two observations indicate that these polymerase mutations affect chromosome stability (Table 3).

Mutator effects of replicative Pol mutations in otherwise wild-type background:
By analogy to the reduced DNA synthesis fidelity of mutant analogues of Klenow fragment Pol, a major objective of this study was to determine if putative active site mutants in eukaryotic replicative polymerases affected mutation rates in vivo. To test this, we measured spontaneous mutation rates in strain 8C-YUNI101 and its four polymerase mutants that were viable as haploids at the five loci that monitor base substitution and frameshift mutations.

For the Y951A Pol and Y831A Pol mutant strains, the mutation rates at all five loci were similar to those seen in the wild-type yeast strain (Table 4). The Y869A Pol mutant exhibited wild-type mutation rates at three loci, but had 3-fold higher rates for canavanine resistance and for His⁺ reversion. DNA sequence analysis of His⁺ revertants of the Y869A Pol strain (Table 5) indicated that the rate of addition of one A·T base pair to the homonucleotide run at the his7-2 mutation site is actually elevated by 22-fold compared to the wild-type strain. The Y708A Pol mutant had a wild-type reversion rate of the trp1-1 and ade2-1 mutation, but mutation rates were elevated 3-fold, 4-fold, and 6-fold, respectively, at the his7-2, ura3-29, and CAN1 loci. DNA sequence analysis of His⁺ revertants of the Y708A Pol strain (Table 5) indicated that the rate of addition of one A·T base pair to the homonucleotide run at the his7-2 mutation site is elevated by 14-fold compared to the wild-type strain. DNA sequence analysis of Ura⁺ revertants of the Y708A Pol strain indicated that the rates of C T and C G substitutions at the ura3-29 site are elevated by 4-fold and 11-fold, respectively, compared to the wild-type strain (Table 6). Thus, two of the polymerase mutants are modest spontaneous mutators for some point mutations in yeast strains that are proficient in DNA mismatch repair. The degree to which the four different polymerase gene mutations affect spontaneous mutation rates is in the same relative order as for the other phenotypes mentioned above (Table 3).

Mutator effects of replicative Pol mutations in mismatch repair-defective strains:
Mutation rates were measured at the five loci in polymerase mutant strains that are also deficient in DNA mismatch repair due to disruption of the PMS1 gene (Table 4B). At the CAN1, trp1-1, and his7-2 loci, mutation rates in a pms1::LEU2 strain that contains normal polymerase genes are substantially higher than are the rates in a mismatch repair-proficient strain (Table 4, compare A and B). This is expected due to lack of repair of mismatches generated by the wild-type replicative polymerases. However, the rate of reversion to Ura⁺ is elevated only two-fold and reversion of Ade⁺ is not elevated in a pms1::LEU2 background. Replication errors in these latter sites may be efficiently corrected by exonucleolytic proofreading and/or they may be inefficiently corrected by PMS1-dependent mismatch repair.

In the pms1::LEU2 strain background, strains harboring each of the four viable polymerase gene mutations had elevated mutation rates at one or more of the five loci examined (Table 4B). The Y951A Pol mutant had an 8-fold higher rate of Ade⁺ reversion, with negligible effects observed at the other loci. The three other polymerase mutant strains all had mutation rates that were elevated by 2- to 180-fold in comparison to the rates in the pms1::LEU2 strain with wild-type polymerases. The Y869A Pol mutant had an 20-fold higher rate of Trp⁺ and Ade⁺ reversion and a 5- and 7-fold higher rate of forward Can^r and frameshift His⁺ mutations, respectively. Note that variability of Trp⁺ and Ade⁺ mutation rates among independent cultures was much higher than that for Ura⁺ and His⁺ reversion, which makes Trp⁺ and Ade⁺ reversions less reliable assays. The pol1-Y869A allele had only a small effect (2-fold) on base substitutions at the ura3-29 locus. The Y831A Pol mutant was an 80-fold mutator for Ade⁺ and an 5-fold mutator for Ura⁺, Trp⁺, His⁺, and canavanine resistance (Can^r). These relatively strong mutator effects contrast with the lack of a mutator effect for this polymerase allele in the wild-type background, suggesting that all errors arising in the pol2-Y831A strain are corrected by mismatch repair. The strain harboring the Y708A Pol allele had the most broad effects on spontaneous mutation. Ade⁺ reversion was increased by 180-fold, Ura⁺ reversion was increased by 50-fold, and Can^r, His⁺, and Trp⁺ mutations were increased by 4- to 7-fold. Sequence analysis of Ura⁺ revertants showed that the rates of three types of base substitutions at the ura3-29 mutation site are elevated by 6- to 79-fold in the Y708A Pol mutant strain, by up to 6-fold in the Y831A Pol mutant strain and by up to 3-fold in the Y869A Pol mutant strain (Table 6). Sequence analysis of His⁺ revertants (Table 5) showed that the rate of addition of an A·T base pair to the homonucleotide run at the his7-2 locus was elevated by 7-fold in the Y869A Pol mutant strain. The rate was elevated by 5-fold in the Y831A Pol mutant strain and by 6-fold in the Y708A Pol mutant strain.

Monitoring reversion at the ura3-29 and his7-2 loci coupled with DNA sequencing of revertants appeared to be a sensitive approach for characterizing the spontaneous mutator phenotypes of the DNA polymerase mutants. Therefore, further analysis of spontaneous mutagenesis was performed using these two markers.

Mutator effects of replicative Pol mutations in diploid strains:
We examined whether the mutator phenotypes observed in haploid strains could be seen when mutant polymerase alleles were present with wild-type polymerase allele in heterozygous diploid strains. For this purpose, we measured the rate of his7-2 reversion in the YUNI200 diploid strain series (Table 7). His⁺ reversion rates in diploid strains homozygous for the polymerase mutations were similar to those observed in haploid strains, and the mutation rate ranking remained the same despite the fact that the pol1-Y869A and pol3-Y708A alleles severely impaired growth in the diploid state (see above). Reversion rates in the mismatch repair-proficient diploids heterozygous for any of three polymerase mutations were undistinguishable to the wild-type diploid (Table 7). The pol1-Y869A and pol3-Y708A were clearly recessive in the mismatch repair-deficient strain. In contrast, the pol2-Y831A allele was semidominant in the pms1 background; the heterozygous diploid strain had a 3.6-fold increase in His⁺ reversion rate over the wild-type diploid strain, as compared with a 7.5-fold increase in homozygous polymerase mutant diploid.

Partial suppression of mutator phenotype of pol3-Y708A mutation by disruption of REV3:
To further investigate the genetic control of spontaneous mutagenesis in the polymerase mutant strains, we used semiquantitative tests to examine the effect of disruption of the REV3, POL4, and RAD30 genes on Can^r mutation and Ura⁺ and His⁺ reversion rates in a set of mismatch repair-proficient and -deficient strains with or without polymerase mutations (data not shown). The only combination of polymerase mutations that showed genetic interaction was REV3 disruption in a pol3-Y708A strain. REV3 encodes the catalytic subunit of DNA polymerase , a specialized polymerase involved in error-prone processing of both endogenous and induced DNA damage in yeast (LAWRENCE 1996 ). Mutation rates were measured at the five loci in the rev3 and the rev3 pol3-Y708A polymerase mutant strains (Table 4). Disruption of REV3 had a 3-fold antimutator effect on the rate of Can^r mutations in a wild-type background, but it had no detectable effect on spontaneous mutation rates at the ura3-29 or his7-2 loci (Table 4A). Disruption of REV3 did not significantly affect the mutation rates in the mismatch repair-deficient pms1 background (Table 4B). However, disruption of REV3 decreased rates of Can^r mutations (6-fold) and Ura⁺ reversions (3-fold) in the pol3-Y708A strain (Table 4A) and decreased rates of Ura⁺ reversions (almost 9-fold) in the pol3-Y708A pms1 strains (Table 4B; Fig 5A). No such reduction in rate of Ura⁺ reversion was seen in strains with combinations of pol3-Y708A and disruptions of either POL4 or RAD30 (Fig 5A). No suppression of His⁺ reversion was observed by any of three specialized polymerase gene disruptions (Fig 5B). Sequence analysis of Ura⁺ revertants in single rev3 mutant showed that C G transversions disappeared from the spectra. C T transitions were approximately as frequent as in wild type and C A transversions were more frequent in a mismatch repair-proficient strain and equal to wild type in a mismatch repair-deficient strain (Table 6). Sequence analysis showed that in a mismatch repair-proficient background, the antimutator effect of the rev3 in the pol3-Y708A strain is mainly due to a decrease in the rate of C G transversions (more than 12-fold comparing to the pol3-Y708A single mutant) and decrease in the rate of C T transitions (7-fold), while C A transversion rates were increased 8-fold (Table 6A). All types of base pair substitutions were decreased in the pol3-Y708A pms1 rev3 strain (Table 6B). These data imply that Pol may contribute to spontaneous mutagenesis at specific loci and for specific types of base pair substitutions in a strain with a mutation in the Pol active site.

Notably, the mutation rates at the ura3-29 locus in the triple pol3-Y708A pms1 rev3 mutant strain remained substantially higher than those observed in the pms1 strain and disruption of REV3 did not suppress the frameshift mutator effect of the pol3-Y708A. Thus, disruption of REV3 only partially suppresses the mutator effects of the pol3-Y708A mutation, implying that this tyrosine to alanine substitution in Pol may directly reduce replication fidelity. Similarly, disruption of REV3 did not suppress the mutator phenotypes of the pol2-Y831A (Pol) or pol1-Y869A (Pol) mutation at the ura3-29 and his7-2 loci, suggesting that these mutations either directly reduce DNA replication fidelity in vivo or their indirect effect is mediated by a Rev3-independent mechanism.

UV sensitivity and mutability of Pol catalytic subunit active site mutants:
The above experiments suggested that the active site mutations in replicative DNA polymerase genes might impair some function in addition to or other than polymerase fidelity. Therefore we investigated the effects of homologous tyrosine to alanine active site changes in the Rev3 Pol, which is not essential for growth. The REV3 gene encodes the catalytic subunit of Pol, which belongs to DNA polymerase family B (Fig 1), and it is necessary for UV mutagenesis. Unlike the situation with the mutant alleles of Pol, -, and - genes, the rev3 mutations did not yield spontaneous mutator effects (data not shown). We then examined UV radiation-induced mutagenesis in these strains. UV light was a potent inducer of Can^r and Ade⁺ mutations in the wild-type strain (Fig 6B and Fig C). Disruption of REV3 decreased cell survival after UV irradiation (Fig 6A) and severely reduced the frequency of UV radiation-induced Can^r (Fig 6B) or Ade⁺ (Fig 6C) mutations. Active site mutations had intermediate effects. The region II tyrosine to alanine mutation rev3-Y980A (Fig 1A) had an effect that was closer to the effect of the REV3 disruption—it lowered UV survival and strongly decreased UV radiation-induced mutagenesis, suggesting severe loss of Pol function. The effects of the region III mutation rev3-Y1093A (Fig 1B) were less pronounced, as this strain was partially UV mutable and its survival after UV irradiation was closer to that of the wild-type strain (Fig 6A). The degree of effects of the two mutations on Rev3 function is in the same relative order as for replicative polymerases.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Structure-function studies have identified conserved residues in family A DNA polymerase active sites that determine the efficiency and fidelity of DNA synthesis. This study uses this information to probe the functions of eukaryotic family B polymerases in the much more complex setting in vivo. The results indicate that two highly conserved tyrosine residues inferred from structural studies to be at the polymerase active sites of yeast DNA polymerases , , , and , the four known family B polymerases in yeast, have important roles in determining cellular growth potential, sensitivity to chemical treatments, and/or genomic stability. In a general sense, the severity of phenotypes conferred by the replicative polymerase mutations correlates with the amount of DNA synthesis these enzymes are thought to perform in cells. Thus, for replicative polymerases, the Pol mutants have the weakest phenotypes, the Pol mutants have intermediate phenotypes, and the Pol mutants have the strongest phenotypes when the three corresponding Pol alleles are compared (Table 3). Exonuclease-deficient Pol is thought to incorporate only a small number of nucleotides to initiate Okazaki fragments and is then replaced by other polymerases that may proofread errors made by Pol, an idea suggested by PERRINO and LOEB 1990 . In addition, some DNA synthesized by Pol could be removed during Okazaki fragment maturation. On the opposite end of this continuum, the Pol mutants are severely defective in several ways, consistent with the idea that Pol performs the bulk of chromosomal DNA replication in eukaryotes (BURGERS 1998 ; WAGA and STILLMAN 1998 ). Each of the mutant replicative polymerase alleles studied here differs from the others in one or more phenotypes (Table 3), indicating that the functions of the two conserved tyrosines are polymerase and residue specific. In addition, homologous mutations in REV3 (Pol) conferred defects in UV mutability, suggesting partial loss of function.

Strains with Tyr to Ala substitutions in conserved region II:
Among the many amino acids that form DNA polymerase active sites, we focused here on two that were anticipated from previous studies to be important for DNA synthesis fidelity. One is a tyrosine in conserved region II of family B enzymes (Fig 1A). The X-ray crystal structures of three family B polymerases reveal that this tyrosine is at the polymerase active site (WANG et al. 1997 ; HOPFNER et al. 1999 ; ZHAO et al. 1999 ). It is in a location suggested by WANG et al. 1997 to be functionally equivalent to a critical glutamate residue in family A enzymes, specifically Glu710 in the Klenow fragment Pol and Glu615 in Klentaq Pol. In the structure of Klentaq Pol complexed with DNA and a dNTP that is poised for catalysis, the side chain of Glu615 closely approaches the 2' position of the deoxyribose of the incoming dNTP in a manner that excludes incorporation of rNTPs during DNA synthesis. Replacement of this side chain with alanine in Klenow fragment Pol reduces catalytic efficiency by 20-fold (POLESKY et al. 1992 ) and strongly reduces discrimination against rNTP incorporation (ASTATKE et al. 1998 ). Similar effects are seen in studies of a homologous mutant of another Pol A family member, T7 DNA Pol (DONLIN and JOHNSON 1994 ). By functional analogy, reduced replication efficiency and/or enhanced rNTP incorporation could explain why the yeast pol3-Y613A (Pol) and pol2-Y645A (Pol) alleles are incompatible with vegetative growth in haploids and are recessive in diploids (Table 3; Fig 3A). Reduced polymerase activity could also explain why the rev3-Y980A strain is almost as UV immutable as the strain with a complete deletion of REV3 (Fig 6). Spores containing the pol2-Y645A (Pol) mutation do produce microcolonies (Fig 3A), indicating minimal retention of some POL2 gene function. However, these microcolonies apparently stop growing and never reach the size of normal colonies. The inability of the pol2-Y645A allele to support vegetative growth appropriately may not simply reflect loss of Pol catalytic function, as was reported for mutations of catalytic residues (DUA et al. 1999 ) or an in frame deletion of the Pol catalytic domain that still permitted slow growth (KESTI et al. 1999 ). Instead, this may be due to some form of aberrant catalysis by Pol, such as incorporation of rNTPs or incorrect dNTPs. Efforts are currently underway to purify the mutant polymerases in order to characterize their biochemical properties.

The haploid strain with the pol1-Y869A substitution is viable. However, it grows slowly, is sensitive to hydroxyurea, and is recessive. This suggests that, although sufficient to carry out the essential function of this polymerase in replication, the catalytic efficiency of this Pol mutant is reduced in comparison to wild-type Pol. Indeed, purified human Pol with a Y865S mutation (where human Tyr865 is homologous to yeast Tyr869) has lower catalytic efficiency than does wild-type Pol (DONG et al. 1993 ). The yeast Pol Y869A mutation also destabilizes the genome, as indicated by reduced chromosome stability and the enhanced rate of single-base additions (22-fold, Table 3 and Table 5) and base substitutions (>20-fold for Trp⁺ and Ade⁺ nonsense mutations reversion, Table 4B). The mutator effects were strongest in a pms1 strain lacking mismatch repair, suggesting that the point mutations may directly reflect replication errors made by the Y869A Pol, at least at specific regions of DNA encoding for tRNA genes, known suppressors of nonsense mutations. This suggestion is in agreement with the 15-fold reduced in vitro fidelity of human Pol with a Y865S mutation (DONG et al. 1993 ) and with the reduced fidelity of Klenow fragment Pol containing the putative functionally analogous E710A substitution (MINNICK et al. 1999 ). However, the effect of the pol1-Y869A on base pair substitutions leading to the ura3-29 reversions was small (3-fold for C A transversions, Table 6B). Klenow fragment Pol with a E710A mutation extends from a T·G mismatch almost 40-fold less efficiently than wild-type enzyme (MINNICK et al. 1999 ) suggesting a possibility that Y869A Pol may pause after making an error, possibly allowing the mismatch to be proofread or extended by another polymerase. It is also possible that the Y869A yeast Pol mutation may alter replication so as to enhance the rate of double strand breaks formation (as implied by the low chromosome stability), thereby indirectly increasing mutagenesis. It was reported earlier that double strand break repair at the TRP1 locus is mutagenic (HOLBECK and STRATHERN 1997 ). Collectively, the observations with the Y869A Pol mutation imply that it may enhance mutagenesis by directly reducing replication fidelity or by contributing to genome instability indirectly through unknown pathway.

Strains with Tyr to Ala substitutions in conserved region III:
The second amino acid residue anticipated to be important for DNA synthesis fidelity is the tyrosine in conserved region III of family B enzymes (Fig 1B). Structural studies (WANG et al. 1997 ; HOPFNER et al. 1999 ; ZHAO et al. 1999 ) indicate that this tyrosine is at the polymerase active site and may be functionally equivalent to a conserved tyrosine in family A enzymes, specifically Tyr766 in Klenow fragment Pol. This amino acid is suggested to contribute to the proper base pairing geometry in the binding pocket for the nascent base pair and/or to act as a chaperone in delivering the dNTP to the binding pocket (reviewed in KUNKEL and BEBENEK 2000 ). Replacement of this side chain in Klenow fragment Pol with alanine reduces its DNA binding affinity, catalytic efficiency, base selectivity, and mismatch extension efficiency (BELL et al. 1997 ; MINNICK et al. 1999 ). The putative functionally analogous mutant of RB69 Pol (a family B member) also has slightly reduced catalytic efficiency (YANG et al. 1999 ) and reduced misincorporation fidelity in vitro and it is a strong spontaneous base substitution mutator in vivo (BEBENEK et al. 2000 ).

Haploid yeast strains with analogous tyrosine to alanine mutations in conserved region III (Fig 1B) of Pol, -, and - are all viable and the homologous mutant of Pol is UV mutable (Fig 6). The pol1-Y951A (Pol) allele had almost no effect on spontaneous mutagenesis, increasing only Ade⁺ reversion in the pms1 background. It may not synthesize DNA at the loci examined, it may not have reduced fidelity, or errors made by this polymerase may be efficiently corrected by proofreading or during Okazaki fragment maturation. The latter possibility is less likely because no genetic interaction of the pol1-Y951A with the proofreading exonuclease mutations pol2-4 and pol3-01 or the flap endonuclease rad27 mutation has yet been found (Y. PAVLOV, unpublished data). In contrast, the pol2-Y831A Pol mutant strain exhibited normal growth and the wild-type sensitivity to MMS treatment and chromosomal stability, but was slightly sensitive to HU and exhibited clear spontaneous base substitution and frameshift mutator phenotypes when the PMS1 was disrupted (Table 4 Table 5 Table 6). These mutator effects were not suppressed by disruption of specialized polymerase genes and the mutator effect at the his7-2 locus is semidominant (Table 7). Given the effects of Tyr to Ala substitutions in Klenow fragment Pol and RB69 Pol mentioned above, these data suggest that the pol2-Y831A allele encodes a mutant Pol whose catalytic efficiency is not severely reduced, but whose base selectivity may be reduced, leading to replication errors that can be corrected by mismatch repair.

Among replicative polymerase mutants that were viable in the haploid state, the most severe phenotypes were exhibited by Y708A Pol mutant (Table 3). A strain with the pol3-Y708A allele grew slowly and the mutation was recessive, indicating that the mutant polymerase does not compete effectively with wild-type Pol. The haploid pol3-Y708A strain was sensitive to hydroxyurea and MMS, suggesting that both DNA replication and DNA repair are defective. Chromosomal stability is low in this strain, as judged by an elevated rate of loss of centromeric plasmids in haploid (our unpublished observations) and by the bipolar mating of homozygous diploid (Fig 4). The pol3-Y708A strain also has strongly elevated rates of spontaneous base substitution and frameshift mutations (Table 4 Table 5 Table 6 Table 7). Larger deletions and more complex mutations in the CAN1 gene were abundant in the pol3-Y708A strain (our unpublished observations). The mutator effects were strongest in the mismatch repair-defective background (Table 4), especially for base substitutions at the ura3-29 and ade2-1 loci (Table 4). The base substitution specificity of the pol3-Y708A Pol mutant is quite different from that of the pol2-Y831A Pol mutant (Table 6). Thus the homologous region III Tyr to Ala changes in DNA polymerases and both enhance spontaneous errors that are subject to mismatch repair, but with quite different specificity. This could reflect a different misinsertion specificity of these mutant enzymes, or a differential ability to proofread errors. The latter could result from altered communication between the polymerase and exonuclease active sites due to the mutation in the polymerase active site as has been suggested for other mutant family B enzymes (see, for example, REHA-KRANTZ and NONAY 1994 ; BEBENEK et al. 2000 ). It is worth noting that the pol3-Y708A and pol2-Y831A alleles encode wild-type active sites for their intrinsic exonucleases. Thus, the mutagenic potential of these mutant polymerases could be underestimated if misinsertions and misalignments are proofread by these exonucleases.

Klenow fragment Pol with a Y766A mutation analogous to the pol3-Y708A Pol allele has reduced capacity to extend a template-primer with a terminal mismatch (CARROLL et al. 1991 ). The Y567A mutation in RB69 Pol, a family B homolog of the pol3-Y708A Pol mutant, also extends terminal mismatches less efficiently than does its wild-type parent (BEBENEK et al. 2000 ). If the pol3-Y708A Pol mutant also extends mismatched termini poorly, then the observed strong base substitution mutator phenotype may reflect low base selectivity by this mutant Pol followed by fast dissociation (and thus escape from proofreading by intrinsic exonuclease) and then by mismatch extension catalyzed by a second polymerase that extends mismatched termini more efficiently. This hypothesis is supported by the observation that disruption of REV3, encoding DNA polymerase that is devoid of proofreading exonuclease, reduces the rate of the pol3-Y708A-dependent spontaneous base substitutions that revert the ura3-29 mutation (Table 4 and Table 6; Fig 5). A major component of this antimutator effect in a mismatch repair-proficient strain is suppression of C G transversions that may arise due to mismatches poorly correctable by proofreading and mismatch repair (see MORRISON and SUGINO 1994 ; LUHR et al. 1998 ). Indeed, the ratios of different base substitutions in a mismatch repair-deficient to a mismatch repair-proficient pol3-Y708 strain are 38 for C T, 275 for C A, and 1.6 for C G (Table 6). Thus, mismatches leading to C A transversions in the pol3-Y708A background at ura3-29 locus are most vulnerable to mismatch repair, while those leading to C G transversions are resistant. From the data of Table 6A we can estimate that rate of C G in pol3-Y708A is decreased by the rev3 deletion at least 12-fold. In a mismatch repair-deficient pol3-Y708A strain two other types of base substitutions are also suppressed by the rev3 (C T 17-fold, C A 5-fold; Table 6B). Yeast Pol is known to efficiently extend mismatched primer termini (LAWRENCE 1996 ; JOHNSON et al. 2000 ), such that loss of this function could explain the observed antimutator effect. Another possibility is that mutant Pol is unable to bypass spontaneous DNA damage and dissociates giving opportunity for Pol-dependent error-prone bypass (see HARFE and JINKS-ROBERTSON 2000 ). The moderate mutator effect of the rev3 mutation for C A transversions in a mismatch repair-proficient pol3-Y08A strain raises a possibility that there is a relatively small proportion (seen only when most errors are corrected by a mismatch repair) of mispairs leading to C A transversions that are processed by Rev in an error-free way.

Note that the suppression of spontaneous mutagenesis in replicative Pol mutants by REV3 disruption is specific in three ways. It is observed for disruption of REV3 but not for disruption of RAD30 (Pol) or POL4 (Fig 5). This suggests that the latter two polymerases may not compete for extending aberrant termini generated by Pol. Suppression by disruption of REV3 is observed specifically for pol3-Y708A-dependent mutagenesis but not for other replicative DNA polymerase mutations. This suggests that either Pol does not compete with mutant Pol or Pol for aberrant termini or that the latter polymerases may not need assistance in extending mismatches. Finally, disruption of REV3 strongly reduces rates of some substitution mutations (for example, those arising in the pol3-Y708 pms1 strain), but not all (for example, arising in other polymerase mutants), and disruption of REV3 does not reduce rates of frameshift mutations. The frameshifts may result from misaligned DNA intermediates with termini that are more easily extended and therefore do not require a specialized polymerase for continued synthesis. The high rate of frameshift mutations and the still higher than normal rate of base substitutions even when the Pol gene is disrupted suggest that the pol3-Y708A mutation at the polymerase active site reduces the base selectivity of the major eukaryotic replicative DNA polymerase.

In the pms1 background, pol3-Y708A stimulated frameshift reversion at his7-2 substantially more than it stimulated base pair substitutions that revert ura3-29 (Table 5B and Table 6B). This implies a strong effect of the pol3-Y708A allele on frameshift mutagenesis in homopolymeric tracts. This may explain why rev3 deletion did not decrease the rate of Can^r forward mutations in the pol3-Y708A pms1 strain (Table 4B). For example, it is well known that most mutations at CAN1 in mismatch repair-defective strains are frameshifts in long homopolymeric tracts (MARSISCHKY et al. 1996 ). If the pol3-Y708A-dependent mutations at can1 are also primarily frameshift mutations in long homopolymeric tracts, they would likely involve misaligned substrates that do not contain a terminal mismatch (as mentioned above). These intermediates may not require extension by pol , and the mutations would therefore not be suppressed by deletion of rev3. Thus when the proportion of frameshifts is high, the CAN1 forward mutation system is rather insensitive to changes in rates of base substitutions.

It was shown earlier that the REV3 disruption does not suppress spontaneous mutagenesis for Can^r in the pol3-01 strain, which is defective in Pol proofreading exonuclease and does not exhibit growth defects (SHCHERBAKOVA et al. 1996 ; DATTA et al. 2000 ). Thus, unedited replication errors arising when the Pol active site is normal result in a Pol-independent increase in mutation rate. On the other hand, the mutator effect of the pol2-1 mutation (the URA3 insertion into the center of POL2 gene), which results in extremely slow growth, is REV3 dependent (SHCHERBAKOVA et al. 1996 ). It is possible that the magnitude of the replication defect conferred by a polymerase mutation determines the extent of Pol participation in replication of undamaged DNA.

Effect of active site mutations in Pol on UV survival and mutagenesis:
Mutation of the region II tyrosine in the REV3 gene yields a strong defect in survival and mutagenesis after UV irradiation, while mutation of the region III tyrosine gives milder effects. This implies that these rev3 missense mutations, which are inferred to be at the active site and reduce the fidelity of other polymerases, actually impair rather than improve the ability of Pol to participate in mutagenic translesion synthesis with UV photoproducts. It may be informative to study homologous missense mutations with a partial defect in Pol function in transgenic mice, since complete deletion of mouse Pol results in early embryonic lethality (BEMARK et al. 2000 ; ESPOSITO et al. 2000 ; WITTSCHIEBEN et al. 2000 ).

Dual mechanism of mutagenesis in replicative DNA Pol mutants:
The magnitude of increase and specificity of base pair substitutions in the ura3-29 locus are different in polymerase , , and mutants, which is consistent with different involvement of each of these Pols in replication. Another notable mutator effect of polymerase mutations in a pms1 background is a strong increase in frameshift mutation rate. The magnitude of the frameshift mutator effect was the same in strains with any of three mutations. We detected almost exclusively +1 frameshifts but we know that Pol mutations lead to elevated levels of -1 frameshifts, too (Y. PAVLOV, unpublished observations). Frameshift mutator effects were also observed for other alleles of POL2 and POL3 in combination with defective mismatch repair, with one allele of POL2 exhibiting a specific preference for +1 frameshift mutagenesis (TRAN et al. 1999 ; KIRCHNER et al. 2000 ). The frameshift mutator effect we report here may reflect reduced frameshift fidelity of these mutant polymerases. At least two models could be considered in this respect, misinsertion slippage (BEBENEK and KUNKEL 1990 ) and melting misalignment (FUJII et al. 1999 ).

Alternatively or in addition, the mutator effects seen here could be indirectly due to loss of coordination at the replication fork, perhaps leading to substitution of one DNA polymerase for another. Such a switch might be more likely to occur under conditions of lowered catalytic efficiency and/or a high level of misinsertions by a mutant replicative DNA polymerase as well documented in E. coli (RANGARAJAN et al. 1997