Characterization of the Repeat-Tract Instability and Mutator Phenotypes Conferred by a Tn3 Insertion in RFC1, the Large Subunit of the Yeast Clamp Loader

Corresponding author: Eric Alani, Section of Genetics and Development, Cornell University, 459 Biotechnology Bldg., Ithaca, NY 14853-2703., eea3{at}cornell.edu (E-mail)

Communicating editor: P. L. FOSTER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The RFC1 gene encodes the large subunit of the yeast clamp loader (RFC) that is a component of eukaryotic DNA polymerase holoenzymes. We identified a mutant allele of RFC1 (rfc1::Tn3) from a large collection of Saccharomyces cerevisiae mutants that were inviable when present in a rad52 null mutation background. Analysis of rfc1::Tn3 strains indicated that they displayed both a mutator and repeat-tract instability phenotype. Strains bearing this allele were characterized in combination with mismatch repair (msh2, pms1), double-strand break repair (rad52), and DNA replication (pol3-01, pol30-52, rth1/rad27) mutations in both forward mutation and repeat-tract instability assays. This analysis indicated that the rfc1::Tn3 allele displays synthetic lethality with pol30, pol3, and rad27 mutations. Measurement of forward mutation frequencies in msh2 rfc1:Tn3 and pms1 rfc1:Tn3 strains indicated that the rfc1::Tn3 mutant displayed a mutation frequency that appeared nearly multiplicative with the mutation frequency exhibited by mismatch-repair mutants. In repeat-tract instability assays, however, the rfc1::Tn3 mutant displayed a tract instability phenotype that appeared epistatic to the phenotype displayed by mismatch-repair mutants. From these data we propose that defects in clamp loader function result in DNA replication errors, a subset of which are acted upon by the mismatch-repair system.

MUTATIONS in genes that are involved in DNA replication and repair often result in chromosomal instabilities such as base pair substitutions and frameshifts as well as insertion, deletion, and rearrangement events (i.e., SCHAAPER and DUNN 1987 ; AGUILERA and KLEIN 1988 ; STRAND et al. 1993 ; MCALEAR et al. 1996 ; TRAN et al. 1996 ; GREEN and JINKS-ROBERTSON 1997 ). Mutations that confer these phenotypes have been directly identified in Escherichia coli and Saccharomyces cerevisiae using a variety of chromosome instability assays (i.e., FEINSTEIN and LOW 1986 ; AGUILERA and KLEIN 1988 ; MICHAELS et al. 1990 ; JEYAPRAKASH et al. 1994 ). Such basic research approaches have also led to a better understanding of the underlying chromosome stability defects that have been observed in patients who suffer from particular inherited diseases. For example, the phenotypes exhibited by mismatch-repair-defective E. coli and S. cerevisiae strains provided evidence indicating that the underlying cause for a large percentage of hereditary nonpolyposis colorectal cancers was a defect in mismatch repair (LEVINSON and GUTMAN 1987 ; STRAND et al. 1993 ; CROUSE 1996 ; reviewed in KOLODNER 1996 ; MODRICH and LAHUE 1996 ).

A major type of chromosomal instability that has been identified in yeast, in bacteria, and in cancer cells is repeat-tract instability (reviewed in CROUSE 1996 ; KOLODNER 1996 ; MODRICH and LAHUE 1996 ; SIA et al. 1997B ). This instability is thought to result mainly from the failure to repair DNA slippage events that occur during the replication of repetitive DNA sequences such as those that contain mono-, di-, tri-, and tetranucleotide repeats. In both yeast and mammalian cells, mutations in mutS and mutL homolog mismatch-repair genes result in a large increase (100- to 10,000-fold) in the rate of repeat-tract instability (STRAND et al. 1993 ; TRAN et al. 1997 ; UMAR et al. 1998 ). This increase is thought to be due to the inability of these mutants to repair small loop insertion/deletion mutations that occur during DNA slippage (reviewed in CROUSE 1996 ; KOLODNER 1996 ; MODRICH and LAHUE 1996 ; SIA et al. 1997B ). In yeast, mutations in DNA replication genes that encode the polymerase processivity factor PCNA (POL30) and the flap endonuclease (RTH1/RAD27) have also been shown to confer an increase in repeat-tract instability at a rate that is similar to that observed in mismatch-repair-defective mutants (JOHNSON et al. 1995 , JOHNSON et al. 1996 ; UMAR et al. 1996 ; KOKOSKA et al. 1998 ). Mutations in other DNA replication genes, such as those that encode the Pol (POL3), and Pol (POL2) DNA polymerases, however, resulted in only modest increases in repeat-tract instability (JOHNSON et al. 1995 ; TRAN et al. 1997 ; KOKOSKA et al. 1998 ). These data have led to the proposal that some DNA replication factors function directly in the mismatch-repair pathway to repair loop insertion/deletions that result from DNA slippage events while others act at the level of DNA replication to prevent the formation of these events (KOKOSKA et al. 1998 ; reviewed in SIA et al. 1997B ).

The above observations, in conjunction with the observation that many cancer cells display repeat-tract instabilities that are unlinked to previously identified repair and replication genes, encouraged us to initiate screens in yeast to identify chromosomal instability mutants (LIU et al. 1995 ). We hoped to identify additional factors that are involved in preventing mutagenic replication errors by preventing DNA slippage events or by facilitating the repair of these slippages through mismatch-repair mechanisms. In one screen we searched for mutants that displayed an increase in the frequency of both repeat-tract insertion/deletion and base pair substitution/frameshift events; unfortunately, this screen only identified mutants that displayed mutations in the previously characterized mutS (MSH2) and mutL (PMS1, MLH1) homolog genes and the RAD27 gene (Y. XIE, L. SCHVANEVELDT and E. ALANI, unpublished data). In a second screen that is the focus of this article, we searched for mutants that were inviable in a double-strand break-repair-deficient (rad52) background and also displayed a mutator phenotype (COUNTER et al. 1997 ). This screen was conducted on the basis of two observations made in E. coli:

1. Certain DNA replication mutants display chromosome instability defects such as chromosomal breakages that are lethal in recombination-deficient backgrounds (MICHEL et al. 1997 ).

2. Strains that lack dam methylase, and are thus defective in strand discrimination during mismatch repair, display a mutator phenotype and are inviable in recombination deficient (recA^-) backgrounds (MCGRAW and MARINUS 1980 ). This inviability, which can be rescued by mutations in the mutS, mutL, and mutH mismatch-repair genes, is thought to be caused by unrepairable double-strand breaks in dam^- recA^- strains that form as the result of MutH incising both template and newly replicated strands (MCGRAW and MARINUS 1980 ; AU et al. 1992 ). Using this second screen we identified and characterized a transposon Tn3 insertion allele of RFC1, a gene that encodes the large subunit of the highly conserved RFC complex that functions in eukaryotic DNA replication and repair. During DNA replication, RFC interacts with the sliding clamp PCNA at the replication fork primer terminus in steps that require adenosine 5'-triphosphate (ATP). Formation of the RFC-PCNA-primer terminus complex then promotes efficient DNA synthesis by both the and DNA polymerases (TSURIMOTO and STILLMAN 1989 ; BURGERS 1991 ; FIEN and STILLMAN 1992 ; STILLMAN 1994 ).

In this study we tested strains bearing the rfc1::Tn3 allele alone or in combination with mismatch repair and DNA replication mutations for defects in chromosome stability. As described below, the rfc1::Tn3 mutation conferred a mutator phenotype that appeared multiplicative with mismatch-repair mutations. In repeat-tract stability assays the rfc1::Tn3 mutation conferred an ~10-fold increase in the frequency of dinucleotide repeat-tract instability that appeared to be epistatic to the phenotype observed in mismatch-repair-defective mutants. Taken together, our data are consistent with the idea that defects in clamp loader function result in DNA replication errors, a subset of which are identified and repaired by the mismatch-repair system.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media and chemicals:
E. coli strains were grown in Luria-Bertani (LB) broth or on LB agar, which was supplemented with 100 µg/ml ampicillin when required (MILLER 1972 ). Yeast strains were grown in either YPD or minimal selective media (ROSE et al. 1990 ). Selective media contained 0.7% yeast nitrogen base, 2% agar, 2% glucose, and 0.09% of a drop-out mix that lacks the amino acid used for selection. Sporulation media (SPM) were prepared as described previously (DETLOFF et al. 1991 ). When required, canavanine was included in minimal selective media at 60 mg/liter (ROSE et al. 1990 ). 5-fluoro-orotic acid (5-FOA) was purchased from U.S. Biologicals and used as described previously (BOEKE et al. 1984 ). When required, methyl-methane sulfonate (MMS; Aldrich Chemical, Milwaukee) was included in YPD media at 0.017% (v/v).

E. coli strains:
DH5 [F' phi80, dLacZ (lacZYA-argF), U169, recA1, endA1, hsdr17 (r^-K, m⁺K), lambda^-, thi1, gyrA, relA1] and KC8 (gammax1486^-, MK12⁺, leuB600, trpC9830, pyrF::Tn5, hisB463, lacx74, StrA, galU, K) were used to amplify and manipulate all plasmids described in this article.

S. cerevisiae strains:
The genotypes of all strains used in these studies are shown in Table 1. With the exception of NKY1068, EAY561, and DNR53, all strains were derived from the isogenic FY strain background (WINSTON et al. 1995 ). DNR53 is also an S288C-derived strain that contains the rfc1::Tn3 allele (BURNS et al. 1994 ; COUNTER et al. 1997 ). This strain also contains a rad52::HIS3 mutation and is viable because it contains an ARS1, CEN4, URA3, plasmid bearing the wild-type RAD52 gene (COUNTER et al. 1997 ). NKY1068 is an SK1-derived strain (KANE and ROTH 1974 ) that was kindly provided by Douglas Bishop and EAY561 is a rfc1::Tn3 derivative of NKY1068.

The msh2::TRP1, msh2::hisG, pms1::hisG, rad52::URA3, rad52::LEU2, and rad27::HIS3 alleles contain complete or nearly compete coding region deletions of their respective genes and were introduced into FY23 and FY86 by single-step transplacement. The primer sequences that were used to make polymerase chain reaction (PCR)-amplified DNA fragments containing the rad27::HIS3 allele were described by TISHKOFF et al. 1997 and the rad52::URA3 and rad52::LEU2 disruption plasmids were kindly provided by Todd Milne and Dennis Livingston, respectively. All of the other disruption plasmids were made in the Alani laboratory and are available upon request. Double mutant combinations of the alleles described in Table 1 were made by standard crosses (ROSE et al. 1990 ). The pol3-01 and pol30-52 alleles were introduced by two-step transplacement (ROSE et al. 1990 ). Vectors used to introduce the pol3-01 (YIPAM26) and pol30-52 (pBL241-52) alleles into the FY strain background were kindly provided by Akio Sugino and Peter Burgers, respectively. The rfc1::Tn3 allele was introduced into the FY and NKY strain backgrounds by single-step transplacement using DNA that had been PCR amplified from DNR53 chromosomal DNA using RFC1 specific primers. The phenotype of rfc1::Tn3 in the different strain backgrounds was identical with respect to mutator phenotype and cold, MMS, and UV sensitivity. All of the alleles that were introduced by single-step transplacement were confirmed by PCR analysis of chromosomal DNA isolated from the transformed strains (primers available upon request). The introduction of the pol30-52 mutation into the FY strain was confirmed by sequencing PCR-amplified DNA fragments containing the POL30 open reading frame. The presence of the pol3-01 mutation was confirmed by digesting the PCR-amplified POL3 gene from candidate strains with EcoRV (recognition site is lost) or BstUI (recognition site is gained). Detailed information about these restriction enzyme digestion protocols is available upon request.

Genetic techniques:
Yeast were transformed with DNA using the lithium acetate method as described by GEITZ and SCHIESTL 1991 . Tetrads were dissected on YPD plates immediately after zymolyase treatment using previously established methods (ROSE et al. 1990 ). In the tetrad analysis described in Table 3, all tetrads that yielded four, three, and sometimes two and one viable spores were examined for relevant genetic markers by PCR, by segregation of a linked marker (i.e., rad27::HIS3), or by phenotype (i.e., mutator phenotype, MMS^s).

View this table: In this window In a new window   Table 2. Median frequency of forward mutations and dinucleotide repeat-tract instability in wild type, rfc1::Tn3, pms1, msh2, rad27, and pol30-52 strains

View this table: In this window In a new window   Table 3. Tetrad analysis of strains containing rfc1::Tn3::LEU2, pol30-52, pol3-01, rad27, rad52, msh2, and pms1 mutations

Mutation frequencies shown in Table 2 were determined by measuring the frequency of forward mutation to canavanine resistance (i.e., REENAN and KOLODNER 1992 ). Repeat-tract instability frequencies (Table 2 and Table 4) were determined by measuring frameshift events that resulted in resistance to 5-FOA in strains containing the plasmid pSH44[(GT)₁₆T-URA3, ARSH4, CEN6, TRP1] (HENDERSON and PETES 1992 ). In both the mutator and repeat-tract instability studies, tested strains were streaked to form single colonies on selective minimal plates containing 2% glucose. Eleven independent colonies were suspended in water and appropriate dilutions were then plated onto minimal media with or without canavanine or 5-FOA. The median frequency of canavanine and 5-FOA resistance was determined for each experiment and the average of three-to-six independent experiments is presented for each strain.

View this table: In this window In a new window   Table 4. Distribution of poly(GT) tract alterations in wild type, rfc1:Tn3::LEU2, msh2, rad27, pol3-t, and pol30 strains

The length of GT repeat tracts was determined by sequencing pSH44-derived plasmids recovered from independently isolated 5-FOA^r colonies (ROSE et al. 1990 ). Plasmids were sequenced using the -40 sequencing primer described by HENDERSON and PETES 1992 . To examine large insertion/deletion alterations in the CAN1 gene, the complete open reading frame of the CAN1 gene was amplified by PCR from chromosomal DNA isolated from independently identified can^r colonies. Amplified DNA was digested with Hph1 and then electrophoresed on a 2% TAE-agarose gel.

Mitotic recombination frequencies were determined by measuring the frequency of His⁺ colonies in wild-type (NKY1068) and rfc1:Tn3 (EAY561) strains bearing a his4X-ADE2-his4B cassette (BISHOP et al. 1992 ). From each strain 13 independent colonies were plated using the appropriate dilutions onto minimal media with or without histidine. The median frequency of His⁺ recombinants was determined.

The genetic data presented in Table 2 and in the text were analyzed using the Mann-Whitney test statistic where P values <0.05 are considered significant (PFAFFENBERGER and PATTERSON 1977 ).

Nucleic acid and protein techniques:
All restriction endonucleases were purchased from New England Biolabs (Beverly, MA) and used according to manufacturers' specifications. Taq and Expand polymerases were purchased from Perkin-Elmer Cetus (Norwalk, CT) and Boehringer Mannheim (Indianapolis), respectively. Plasmid DNA was isolated by alkaline lysis and all DNA manipulations were performed as described previously (MANIATIS et al. 1982 ). Yeast chromosomal DNA was prepared as described by HOLM et al. 1986 . Preparations were made from 5-ml yeast cultures grown to saturation in YPD. The purified chromosomal DNA was dissolved in 50 µl of double-distilled water and stored at -20° prior to use.

Polymerase chain reaction (PCR) was performed as described previously (SAIKI et al. 1985 ) and amplification conditions and primer sequences for the different reactions are available upon request. The basic reaction was performed for 30 cycles using a denaturation step of 30 sec at 94°, an annealing step of 30 sec at 58°, and a polymerization step of 2 min at 72°. Reactions were performed using Taq polymerase in 25 µl with 5 pmol of each primer and ~1 µg of yeast DNA. The DNA primer synthesis and DNA sequencing were performed at the Cornell Biotechnology Analytical/Synthesis facility.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Identification of a rfc1 allele containing a Tn3::LEU2 insertion:
We screened strains mutagenized with Tn3 transposon insertions for those that were inviable in a double-strand break-repair-deficient rad52 null background (BURNS et al. 1994 ; HOWELL et al. 1994 ; COUNTER et al. 1997 ). A collection of 836 yeast mutants that were identified from ~300,000 colonies was examined for mutator phenotypes on canavanine plates and a single candidate (DNR53, Table 1) was identified (ROSE et al. 1990 ; COUNTER et al. 1997 ). Sequencing of the DNA that flanked the Tn3 insertion in this candidate revealed that the Tn3 element was located at bp 756 in the RFC1 open reading frame (ORF) between homology boxes I (ligase homology domain) and II (Figure 1; CULLMANN et al. 1995 ). The allele created by the insertion is referred to as rfc1::Tn3. Previously, HOWELL et al. 1994 showed that the RFC1 gene product is required for cell viability; however, plasmids containing a deletion variant of the RFC1 gene that lacks the entire aminoterminal ligase homology domain (1-273) can weakly complement the cold-sensitive phenotype of rfc1-1 mutants (HOWELL et al. 1994 ; Figure 1). Consistent with this observation was the finding that the ligase homology domain of the human homolog of RFC1 is not required for PCNA interactions or replication functions in vitro (UHLMANN et al. 1997 ). On the basis of this information we hypothesize that in rfc1::Tn3 strains a cryptic promoter within the Tn3 element is driving expression of an aminoterminal-truncated version of the RFC1.

View larger version (7K): In this window In a new window Download PPT slide   Figure 1. The rfc1::Tn3 allele contains a mini-Tn3:: LACZ::LEU2 (BURNS et al. 1994 ) transposon insertion at bp 756 in the RFC1 open reading frame. The transposon insertion in RFC1 is located between box I (ligase homology domain) and II in the RFC1 ORF. Boxes II–VIII are defined as containing short motifs that are conserved among the five subunits of the yeast and human clamp loader complex. Boxes III and V contain conserved sequences characteristic of nucleotide-binding proteins; the functions of the other boxes are unknown (CULLMANN et al. 1995 ).

Characterization of the rfc1::Tn3 phenotype:
The rfc1::Tn3 allele that was identified in DNR53 was introduced into the FY and SK-1 strain backgrounds (Table 1; MATERIALS AND METHODS) and tested in DNA repair and mutator assays. This strain displayed a 19-fold increase in the frequency of forward mutations to can^r (P = 0.034), was sensitive to UV and the alkylating agent MMS, was cold sensitive for growth at 14°, and displayed a 9.5-fold higher median frequency of mitotic His⁺ recombinants compared to wild type (4.3 x 10^-4 for wild type vs. 4.1 x 10^-3 for rfc1::Tn3) in an intrachromosomal gene conversion assay (Table 2; MATERIALS AND METHODS; data not shown; BISHOP et al. 1992 ). The replication and repair defects that were observed in rfc1::Tn3 strains were similar to those described for rfc1 conditional mutants (HOWELL et al. 1994 ; MCALEAR et al. 1996 ).

The rfc1::Tn3 mutation is synthetically lethal with rad52, rad27, pol30-52, and pol3-01 mutations:
Several lines of genetic evidence have begun to reveal the complex interplay of replication and repair functions in both the generation and repair of mutagenic replication errors (misincorporation or repeat-tract insertion/deletion events). Studies of MORRISON et al. 1993 revealed that a polymerase proofreading mutant (pol3-01) displayed multiplicative mutator defects with mismatch-repair mutants, suggesting that these two functions act in series in the repair of replication errors. The PCNA (POL30) and RAD27 replication factor genes have also been implicated in mismatch avoidance and/or correction, as rad27 and certain pol30 (pol30-52 and pol30-104) mutants display a mutator/slippage phenotype that is similar to that observed in msh2 mutants (STRAND et al. 1993 ; JOHNSON et al. 1996 ; UMAR et al. 1996 ; KOKOSKA et al. 1998 ). Finally, genetic analyses of RFC1 and the flap endonuclease RAD27 have revealed that alleles of both are lethal in a rad52 null background and are also synthetically lethal with each other (Table 3; TISHKOFF et al. 1997 ; MERRILL and HOLM 1998 ). This information, in conjunction with recent data from the Holm laboratory, where they showed that rfc1 mutants accumulate small single-stranded DNA fragments (MERRILL and HOLM 1998 ), supports the idea that rad27 and rfc alleles are both defective in the maturation of Okazaki fragments.

The above information encouraged us to explore the interplay of rfc1::Tn3 with mutants in these unlinked replication and repair functions. Haploid strains containing the rfc1::Tn3, rad27, rad52, msh2, pms1, pol3-01, and pol30-52 mutations were mated to each other and tetrads from the resulting diploids were examined for spore viability, segregation of markers, and mutator and repeat-tract instability phenotypes (Table 2 Table 3 Table 4). Double mutant combinations (i.e., crosses 6–9, Table 3) were classified as viable on the basis of the following: (1) The majority of tetrads dissected contained four viable spores (91–99% spore viability). (2) Genotyping analysis of several four-spore viable tetrads from each cross resulted in the identification of double mutant strains that were analyzed in the mutator and repeat-tract instability assays described below. Double mutant combinations (i.e., crosses 1–5, Table 3) were classified as inviable on the the basis of the following: (1) The segregation patterns of tetrads bearing four (PD), three (TT), and two (NPD) viable spores fit the expected pattern for double mutant lethality in the case where two genes are segregating independently of each other (1 PD: 4 TT: 1 NPD). This pattern is also manifested in reduced spore viability. (2) The inviability of double mutant combinations was confirmed by genotyping all spore clones in tetrads containing four and three viable spores, and in some cases tetrads that contained two or one viable spores were genotyped. In cases of synthetic lethality, no spore clones were identified that contained both mutations.

In control crosses, genotyping and spore viability analysis demonstrated that rfc1::Tn3 rad52, rad27 rad52, and pol30-52 rad52 double mutants were inviable (data not shown); this was expected because rad27 rad52 strains were previously shown to be inviable (TISHKOFF et al. 1997 ), and different mutant alleles of the RFC1 (rfc1-1; MERRILL and HOLM 1998 ) and POL30 (pol30-104; MERRILL and HOLM 1998 ) genes were shown to be synthetically lethal with rad52 null mutations. pol3-01 rad52 double mutants were found to be viable (data not shown); this result was also expected because previous analysis indicated that the rad52 null mutation did not exhibit synthetic lethality with mutations in the pol, pol, and pol polymerase genes (MERRILL and HOLM 1998 ). rfc1::Tn3 strains were also mated to strains bearing the pol30-52, rad27, and pol3-01 mutations. As shown in Table 3, rfc1::Tn3 pol30-52 (Cross 1), rfc1::Tn3 rad27::hisG (Cross 2), and rfc1::Tn3 pol3-01 (Cross 3) double mutants were inviable because poor spore viability (59–78%) was observed in tetrad analysis and no spore clones were obtained that contained both mutations.

As shown in Table 2 and Table 3, rfc1::Tn3 msh2::hisG (Cross 7) and rfc1::Tn3 pms1::hisG (Cross 8) double mutants were viable and displayed the MMS^s and cold^s phenotype conferred by the rfc1::Tn3 allele and a mutator phenotype that was nearly equivalent to the product of the mutator frequencies of the individual mutants. The viability of these double mutants encouraged us to test whether, analogous to the suppression of dam^- recA^- lethality by mutS^- mutations, a msh2 mutation could suppress the lethality observed in rfc1::Tn3 rad52 double mutants (MCGRAW and MARINUS 1980 ). In crosses between a rfc1::Tn3 strain and rad52 and msh2 rad52 strains (Table 3, Cross 5; data not shown), no spore clones were recovered that contained both (rfc1::Tn3 and rad52 ) or all three (msh2 rfc1::Tn3, and rad52) mutations. This observation was also confirmed by showing that a msh2 derivative of DNR53 (relevant genotype rfc1::Tn3, msh2, rad52, pRAD52 ARS-CEN URA3) was not viable on 5-FOA media that selected for the loss of the pRAD52 plasmid (data not shown).

The rfc1::Tn3 mutant displays a repeat-tract instability phenotype:
The phenotype of the rfc1::Tn3 allele, as well as previous studies indicating that mismatch repair and DNA replication mutants displayed an increased frequency of repeat-tract instability, encouraged us to test whether the rfc1::Tn3 mutation confers a similar defect (Table 2; STRAND et al. 1993 ; JOHNSON et al. 1996 ; KOLODNER 1996 ; UMAR et al. 1996 ; TRAN et al. 1997 ). We measured the frequency of tract instability in an assay that detects frameshift events resulting in resistance to 5-FOA. These tests were performed in FY23- or FY86-derived strains containing the GT repeat-tract plasmid pSH44 [(GT)₁₆T-URA3, ARSH4, CEN6, TRP1] (HENDERSON and PETES 1992 ). As shown in Table 2, the rfc1::Tn3 allele displayed a moderate repeat-tract instability phenotype (10-fold increased, P = 0.049) that was lower than that observed in mismatch-repair mutants but was similar to that observed in strains bearing the DNA polymerase mutations pol3-01 and pol3-t (TRAN et al. 1997 ; KOKOSKA et al. 1998 ).

pSH44-derived plasmids obtained from independent 5-FOA^r rfc1::Tn3 colonies were sequenced to examine the repeat-tract sequence changes that had occurred (Table 4). The majority (20/34) of tract alterations in rfc1::Tn3 strains were 1-repeat insertion mutations. The remaining alterations comprised one group (6/34) consisting of 1- or 2-repeat insertion/deletion mutations and another group (8/34) consisting of larger 7- to 10-repeat insertion/deletion mutations. This spectrum of tract alterations was somewhat similar to that observed in wild-type and rad27 strains in that the majority of tract alterations in all three backgrounds were single repeat insertion mutations. However, compared to the rad27 strain, the rfc1::Tn3 strain displayed a higher number of larger tract insertions/deletions (8/34 for rfc1::Tn3 vs. 1/35 for rad27).

The spectrum of repeat-tract insertion/deletion events in the rfc1::Tn3 strain differed from that observed in mismatch repair (msh2) and other DNA replication (pol30-104 and pol3-t) defective strains (Table 4 and JOHNSON et al. 1995 , JOHNSON et al. 1996 ; KOKOSKA et al. 1998 ). In msh2 strains only single repeat insertion/deletions were observed, with the majority of these events consisting of deletions (JOHNSON et al. 1996 ). In pol30-104 strains, which display a mutator and repeat-tract instability phenotype similar to that observed in mismatch-repair mutants, the vast majority (56/58) of tract alterations were one- or two-repeat insertion/deletion mutations, with a similar number of insertions and deletions (JOHNSON et al. 1996 ). In pol3-t strains, which contain a mutation in polymerase that is thought to reduce the rate of DNA elongation, there was an approximately equal distribution of small and large repeat insertion/deletion mutations and a smaller number of alterations that were presumably not due to repeat-tract instability (KOKOSKA et al. 1998 ).

Recent analysis of rad27 strains revealed that, in addition to repeat-tract length instability, these strains also displayed a high frequency of insertion/deletion events at the CAN1 and LYS2 loci (>14 bp, the majority of which were duplications) (JOHNSON et al. 1995 ; TISHKOFF et al. 1997 ; KOKOSKA et al. 1998 ). The similarity in the spectrum of tract instability events in rfc1::Tn3 and rad27 mutants and the fact that rfc1::Tn3 rad27, rfc1::Tn3 rad52 and rad27 rad52 strains are inviable (Table 3 and TISHKOFF et al. 1997 ) encouraged us to examine whether similar types of chromosomal rearrangements could be detected in rfc1::Tn3 strains. We examined rfc1::Tn3, rad27::HIS3, and pol30-52 strains for the presence of large insertion/deletion mutations in the CAN1 locus (Figure 2). As predicted, a difference in the size of at least one CAN1-derived fragment was observed in CAN1 genes amplified from 9 out of 10 can^r rad27 strains (Figure 2, lanes 1–10). However, no changes were observed in the size of CAN1 gene fragments amplified from 10 can^r rfc1::Tn3 (Figure 2, Lanes 12–21) or 8 can^r pol30-52 strains (data not shown). This finding is also consistent with results of MCALEAR et al. 1996 , which showed that the rfc1-1 mutation conferred a mutator phenotype that resulted principally from an elevated frequency of base pair substitution mutations.

View larger version (40K): In this window In a new window Download PPT slide   Figure 2. The rfc1::Tn3 mutation does not result in large insertion/deletion mutations in the CAN1 gene. Independent canr colonies were obtained from rad27::HIS3 (EAY 545) and rfc1::Tn3 (EAY 547) strains, and a DNA fragment containing the CAN1 open reading frame was PCR amplified from each of these strains and incubated with HphI. Restriction enzyme-digested DNA fragments were electrophoresed in a 2% TAE-agarose gel. Lanes marked M contain a DNA marker and lanes 11 and 22 display DNA fragments from wild-type cans strains. Digestion of the amplified CAN1 DNA with HphI resulted in 490-, 411-, 314-, 252-, 249-, and 207-bp fragments that could be detected by gel electrophoresis. Two smaller bands of 87 and 46 bp could not be detected. Lanes 1–10 and 12–21 display HphI-digested CAN1 DNA from canr rad27::HIS3 and rfc1::Tn3 strains, respectively.

Double mutant analysis indicated a synergistic relationship between rfc1 and mismatch-repair mutations:
To test whether RFC1 is required during mismatch repair, we examined the mutation frequency of rfc1::Tn3 strains in combination with mismatch repair and other replication mutations in both forward mutation and tract instability assays. As shown in Table 2, in the forward mutation assay, the frequency of mutations in rfc1::Tn3 msh2 (469-fold increase) and rfc1::Tn3 pms1 (422-fold increase) double mutant strains appeared to nearly equal the product of the mutation frequencies of the individual mutations (msh2, 50-fold increase; pms1, 55-fold increase; rfc1::Tn3, 19-fold increase). While results in the forward mutation assay indicated a nearly multiplicative relationship for defects in the clamp loader and mismatch-repair genes, the results from the tract instability assay were less clear. As shown in Table 2, frequency of tract instability in msh2, pms1, and pol30-52 strains (275- to 500-fold) was much higher than in rfc1::Tn3 strains (10-fold). The frequencies of tract instability in rfc1::Tn3 msh2 and rfc1::Tn3 pms1 double mutants, however, were similar to those observed in msh2 or pms1 strains (rfc1::Tn3 msh2 vs. msh2, P = 0.25; rfc1::Tn3 pms1 vs. pms1, P = 0.66).

Double mutant analysis also indicated that rfc1::Tn3 pms1 and rfc1::Tn3 msh2 mutants were viable but rfc1::Tn3 pol30-52 double mutants were inviable (Table 3). We were interested in understanding this result because pol30-52 mutants display a strong mismatch-repair defect that appears to be epistatic to the defect observed in msh2 and pms1 mutants (UMAR et al. 1996 ). However, given the known role of PCNA and RFC in DNA replication, this result suggests that pol30-52 strains display defects in cellular functions such as DNA replication that are in addition to or are different from those involving mismatch repair. Analysis of msh2 pol30-52 double mutants in the forward mutation and repeat-tract instability assays also supported this conclusion. In both assays, the msh2 pol30-52 mutant displayed a mutator and repeat-tract instability phenotype that appeared to be roughly additive when compared to the mutator phenotype observed for each of the single mutations (Table 2). This observation supports the idea that the mutator phenotype exhibited by pol30-52 and msh2 mutants reflects the action of gene products functioning in parallel, noncompeting pathways (HAYNES and KUNZ 1981 ; MORRISON et al. 1993 ). In such a scenario, the pol30-52 mutation confers replication errors as the result of defects in both mismatch repair and DNA replication. The finding that the pol30-52 mutation, unlike mismatch-repair mutations, confers sensitivity to DNA-damaging agents and, like rad27::HIS3 and rfc1::Tn3, is synthetically lethal with rad52 null mutations, supports this idea; moreover, pol30-52 is defective in homotrimeric interactions and is also defective in in vitro DNA replication reactions (Table 3; AYYAGARI et al. 1995 ). Further support is provided from the work of KOKOSKA et al. 1999 (accompanying article) in which they analyzed the effect of the pol30-52 mutation on micro- and minisatellite instability and found that, in addition to conferring a defect in mismatch repair, the pol30-52 mutation conferred a minisatellite instability phenotype. They hypothesized that this phenotype was the result of the pol30-52 mutation affecting DNA replication by increasing the rate of DNA polymerase slippage.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We identified the rfc1::Tn3 allele in a screen for mutations that are lethal in a rad52 null background. This analysis indicated the rfc1::Tn3 allele conferred a mutator phenotype, an elevated recombination frequency, sensitivity to UV and MMS, cold sensitivity, and synthetic lethality with rad52, rad27, and pol30 mutations. The phenotypes conferred by the rfc1::Tn3 allele were similar to those reported for previously characterized rfc1 alleles (MOIR et al. 1982 ; HOWELL et al. 1994 ; MCALEAR et al. 1994 ). In addition, we showed that rfc1::Tn3 displays a repeat-tract instability phenotype. This observation encouraged us to test genetic interactions between the rfc1::Tn3 mutation and mutations in other replication (RAD27, POL3) and repair (MSH2, PMS1) genes.

In the canavanine mutator assay, a nearly multiplicative effect on mutation frequency was observed when the rfc1::Tn3 mutation was analyzed in combination with msh2 and pms1 mutations. A multiplicative relationship was previously observed in pol3-01 pms1 double mutants; this observation led MORRISON et al. 1993 to propose that DNA replication errors resulting from defects in polymerase proofreading functions are repaired by the mismatch-repair pathway. Because our data displayed a relationship that was almost, but not exactly, multiplicative, we cannot distinguish whether RFC and mismatch-repair functions act in series in a single repair pathway, as proposed for pol3-01 and pms1 double mutants in MORRISON et al. 1993 , or act in distinct DNA repair pathways that compete for the same substrates (HAYNES and KUNZ 1981 ). On the basis of the known function of RFC as a clamp loader in DNA replication, we favor the idea that defects in the clamp loader result in replication errors that are acted upon by mismatch repair (acting in series; MORRISON et al. 1993 ). The types of mutations found in a rfc1-1 strain that confers repair and replication defects similar to those observed in rfc1::Tn3 strains provides further support for this idea (MCALEAR et al. 1996 ). The rfc1-1 mutation was shown to confer an increase in base pair substitutions that are likely to have formed from base pair mismatches that are substrates for mismatch repair (reviewed in KOLODNER 1996 ).

A similar relationship between mismatch repair and RFC functions was not observed in the repeat-tract instability assay because the rfc1::Tn3 msh2 or rfc1::Tn3 pms1 double mutants displayed a mutator phenotype that was indistinguishable from that observed in msh2 or pms1 single mutants. The different phenotypes in these two assays were not unexpected considering that RFC1, MSH2, and PMS1 are likely to be functioning in multisubunit DNA replication and repair machines and it is difficult to determine which effects are direct and which are indirect. Three possible explanations for these differences are as follows:

1. Repeat-tract instability in msh2 and pms1 mutants, as measured by the (GT)₁₄-T-URA3 detection assay, is already occurring at a saturating level and so an increase in these events could not be detected in double mutant combinations. We believe that this is not the case because higher frequencies of repeat-tract instabilities have been observed in pol30-52 msh2 strains and even higher frequencies have been reported in msh2 strains containing mononucleotide repeat tracts. (Table 2; SIA et al. 1997A ).

2. The increase in repeat-tract instability in rfc1::Tn3 strains resulted not from DNA polymerase slippage events but from an increase in unequal sister chromatid exchanges (STRAND et al. 1993 ). The fact that the rfc1::Tn3 allele confers a hyper-recombination phenotype provides support for this idea. In such a scenario one would expect the repeat-tract instability observed in rfc1::Tn3 strains to be dependent on RAD52 function. Unfortunately, this hypothesis cannot be tested as rfc1::Tn3 mutants are lethal in rad52 null backgrounds.

3. The repeat-tract instability phenotype in rfc1::Tn3 strains could result from an increase in DNA slippage events that are not recognized by, or occur independently of, the mismatch-repair system. The fact that ~25% of the tract alterations in rfc1::Tn3 strains were larger in size (>14 bp) than would be expected to be repaired by the mismatch-repair pathway provides some support for this idea (SIA et al. 1997A ), as do recent observations suggesting that the pol30-52 mutation increases the rate of repeat-tract instability through mechanisms that appear independent of mismatch repair [ Table 2; KOKOSKA et al. 1999 ].

Synthetic lethality analysis and chromosome instability assays indicate that the rfc1, pol30, and rad27 mutations interact genetically but display unique chromosome instability phenotypes:
Genetic analysis of rfc1, pol30, and rad27 mutations indicated that they were all synthetically lethal with mutations in the RAD52 recombinational repair pathway and that rfc1 pol30 and rfc1 rad27 double mutants were also inviable (MERRILL and HOLM 1998 ). Additional analysis also showed that mutations in all three of these genes resulted in a mutator phenotype as well as sensitivity to DNA-damaging agents. One hypothesis that is consistent with the above data is that mutations in each of these genes result in the stalling of the replication fork, thus leading to the formation of double-strand breaks that are repaired by the RAD52 double-strand break repair system. This hypothesis is partly based on the findings in E. coli that mutations that disrupt replication fork movement result in the formation of double-strand breaks that are lethal in recombination-deficient backgrounds and the observation that rfc1 mutants grow slowly and are delayed in progressing through S phase (HOWELL et al. 1994 ; MCALEAR et al. 1996 ; MICHEL et al. 1997 ). In addition, recent studies by the Holm laboratory (MERRILL and HOLM 1998 ) indicated that rad27, pol30, and rfc1 mutants accumulate small single-stranded DNA fragments during DNA replication in vivo, suggesting that they are defective in Okazaki fragment maturation.

On the basis of the phenotypes of the mutants described above and the observation that Pol30p physically interacts with the RFC and Rad27p, one might have expected the spectrum of chromosomal instability defects in these mutants to be similar (BURGERS 1991 ; LI et al. 1995 ; UHLMANN et al. 1997 ). However, as outlined below, the chromosome instability profiles of the pol30, rad27, and rfc1 mutants are different. (1) The frequency of repeat-tract instability in rfc1::Tn3 strains was much lower than was observed in rad27 or pol30-52 mutants (Table 2; JOHNSON et al. 1995 , JOHNSON et al. 1996 ; UMAR et al. 1996 ; TISHKOFF et al. 1997 ; KOKOSKA et al. 1998 ; MERRILL and HOLM 1998 ) and the spectrum of repeat-tract changes was unique for each strain (Table 4). In an independent study, P. GREENWELL, R. J. KOKOSKA and T. D. PETES (personal communication) found that the rfc1-1 allele, which displays a phenotype similar to the rfc1::Tn3 allele, confers a repeat-tract instability phenotype that is independent of the size of the repeat unit; for each repeat unit examined (1–20 bp), the rfc1-1 mutation conferred an ~10-fold increase in the rate of repeat-tract instability. This spectrum of changes was very different from that observed in rad27 and pol3-t mutants where the effect of these mutations on repeat-tract instability was dependent on the size of the repeat unit (KOKOSKA et al. 1998 ). (2) Larger DNA rearrangements, as analyzed by DNA sequencing or by restriction digest analysis of the CAN1 gene, occurred at high frequency in rad27 strains but were not observed in rfc1::Tn3 or pol30-52 strains (Figure 2; TISHKOFF et al. 1997 ). In summary, while each of these mutations appears to decrease the efficiency of DNA replication in ways that are likely to stall the DNA replication fork, their overall effect on chromosomal instability appears to be unique for each mutation and is not well understood at this time.

In this study we observed that the pol3-01 mutation, which causes a defect in the polymerase- exonuclease proofreading function, was synthetically lethal with the rfc1::Tn3 mutation (Table 3). Previous analysis by MORRISON et al. 1993 showed that haploid pol3-01 pms1 double mutants are inviable because of the accumulation of a catastrophic number of mutations. While the inviability of pol3-01 rfc1::Tn3 mutants might be explained in this manner, we believe that this is not the case because msh2, pms1, and pol3-01 single mutants all display similar mutation frequencies and rfc1::Tn3 msh2 and rfc1::Tn3 pms1 double mutants are viable (MORRISON et al. 1993 ; KOLODNER 1996 ; UMAR et al. 1996 ). A similar conclusion was reached by KOKOSKA et al. 1998 to explain why pol3-01 rad27 double mutants are inviable. They hypothesized that the inviability of pol3-01 rad27 mutants was not due to mutational load because diploids homozygous for both mutations were also inviable. In cases where mutational load was suspected as the cause of haploid inviability, diploids homozygous for the mutational load mutations were found to be viable, presumably because recessive lethal mutations occur less frequently in diploid cells (MORRISON et al. 1993 ). On the basis of the observation that pol3-01 rad27 double mutants are inviable and the fact that the RAD27 gene product is required for Okazaki fragment processing, KOKOSKA et al. 1998 proposed that the proofreading exonuclease function of POL3 was somehow required for Okazaki fragment processing. The fact that rfc1, pol30, and rad27 mutants all display Okazaki fragment processing defects and rfc1 and rad27 mutations are inviable in a pol3-01 background is consistent with this idea (MERRILL and HOLM 1998 ).

RFC is unlikely to play a direct role in mismatch repair:
A major question that remains to be answered in eukaryotic mismatch repair is how strand discrimination is accomplished so that the newly replicated strand containing the replication error is removed. This question is of interest because in the yeast genome no strand discrimination homologs based on the dam/mutH system of E. coli have been identified (reviewed in KOLODNER 1996 ). As outlined in the Introduction, the rfc1::Tn3 mutation was isolated in a screen designed to identify strand discrimination candidates. However, double mutant analysis of the mismatch-repair mutations msh2 and pms1 and the rfc1::Tn3 mutation indicated that the RFC was unlikely to play a role in mismatch repair because the mutator phenotype of rfc1 was not epistatic to, but was in series with, the mutator phenotypes conferred by msh2 and pms1 mutations. The repeat-tract instability phenotype conferred by the rfc1::Tn3 mutation was weak relative to the mutator phenotype as judged by canavanine papillation assays and in comparison with the repeat-tract instability phenotype observed in msh2 and pol30-52 strains (STRAND et al. 1993 ; UMAR et al. 1996 ). A mutant defective in strand discrimination would be expected to function downstream of mismatch recognition steps and would therefore be expected to display an equally strong mutator and tract instability phenotype (reviewed in KOLODNER 1996 ). In addition, the spectrum of repeat-tract insertion and deletion events observed in rfc1::Tn3 strains was different from that observed in msh2 strains because a large number of tract alterations in rfc1::Tn3 strains involved >14 bp insertion/deletions: such alterations were not observed in mismatch-repair mutants (Table 4; STRAND et al. 1993 ; JOHNSON et al. 1995 ). Finally, unlike in E. coli where dam^- recA^- double mutant lethality can be suppressed by a mutation in mutS, in the analogous situation in yeast the msh2 mutation was unable to suppress the lethality of rfc1::Tn3 rad52 double mutants (see RESULTS; MCGRAW and MARINUS 1980 ). This observation provides further support that the RFC is not playing a strand discrimination role analogous to that of dam methylase.

As described in the Introduction, we performed another screen aimed at identifying new mutations in mismatch-repair genes. We screened 70,000 cells mutagenized with ultraviolet light for those that displayed both a repeat-tract instability and a mutator phenotype. The 51 mutants that were identified all contained mutations in either the MLH1, PMS1, or MSH2 genes. In a subsequent screen involving 220,000 UV-mutagenized cells that was designed to avoid detection of these three genes, only mutations in RAD27 were identified (Y. XIE, L. SCHVANEVELDT and E. ALANI, unpublished data). Why were mutations in genes that are likely to play a role in mismatch repair (i.e., helicases, single-strand binding proteins, and exonucleases) not identified? One possibility is that mismatch-repair functions are overlapping or are redundant because studies in E. coli have suggested that at least three exonucleases participate in mismatch repair (HARRIS et al. 1998 ). Another possibility is that any remaining uncharacterized mismatch-repair factors also play essential roles during DNA replication: in such a scenario, a transposon mutagenesis scheme might not allow for the identification of an essential mismatch-repair protein such as a single-strand binding protein or a DNA helicase, while a genome-wide UV mutagenesis might not have been efficient enough to detect mutations that are essential for viability but are specifically defective in mismatch repair. It is unlikely, for example, that we would have detected mutations in the POL30 gene, the product of which is a strand discrimination candidate, using Tn3 or UV mutagenesis screens (i.e., pol30-52) because POL30 is essential for viability and only rare mutations in POL30 that were found by targeted mutagenesis display both a strong tract instability and mutator phenotype (AYYAGARI et al. 1995 ). This information suggests that fractionation of cell extracts that catalyze mismatch repair in vitro might be a more effective way to identify new mismatch-repair components.

	ACKNOWLEDGMENTS

We thank Elizabeth Evans, Todd Milne, Lori Schvaneveldt, Tanya Sokolsky, Daniel Smith, and Barbara Studamire for providing advice, reagents, and/or technical assistance, Elizabeth Evans, Tom Petes, Daniel Smith, Tanya Sokolsky, and Barbara Studamire for helpful discussions and comments on the manuscript, and Tom Petes and Michael Liskay for sharing unpublished data. E.A. and Y.X. were supported by National Institutes of Health grant GM53085 and U.S. Department of Agriculture Hatch Grant NYC-186424.

Manuscript received August 3, 1998; Accepted for publication October 22, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AGUILERA, A. and H. L. KLEIN, 1988  Genetic control of intrachromosomal recombination in Saccharomyces cerevisiae. I. Isolation and genetic characterization of hyper-recombination mutations. Genetics 119:779-790[Abstract/Free Full Text].

AU, K. G., K. WELSH, and P. MODRICH, 1992  Initiation of methyl-directed mismatch repair. J. Biol. Chem. 267:12142-12148[Abstract/Free Full Text].

AYYAGARI, R., K. J. IMPELLIZZERI, B. L. YODER, S. L. GARY, and P. M. BURGERS, 1995  A mutational analysis of the yeast proliferating cell nuclear antigen indicates distinct roles in DNA replication and DNA repair. Mol. Cell. Biol. 15:4420-4429[Abstract].

BISHOP, D. K., D. PARK, L. XU, and N. KLECKNER, 1992  DMC1: a meiosis-specific yeast homolog of E. coli recA required for recombination, synaptonemal complex formation, and cell cycle progression. Cell 69:439-456[Medline].

BOEKE, J. D., F. LACROUTE, and G. R. FINK, 1984  A positive selection for mutants lacking orotidine-5'-phosphate decarboxylase activity in yeast: 5-fluoro-orotic acid resistance. Mol. Gen. Genet. 197:345-346[Medline].

BURGERS, P. M., 1991  Saccharomyces cerevisiae replication factor C. II. Formation and activity of complexes with the proliferating cell nuclear antigen and with DNA polymerases delta and epsilon. J. Biol. Chem. 266:22698-22706[Abstract/Free Full Text].

BURNS, N., B. GRIMWADE, P. B. ROSS-MACDONALD, E. Y. CHOI, and K. FINBERG et al., 1994  Large-scale analysis of gene expression, protein localization, and gene disruption in Saccharomyces cerevisiae.. Genes Dev. 8:1087-1105[Abstract/Free Full Text].

COUNTER, C. M., M. MEYERSON, E. N. EATON, and R. A. WEINBERG, 1997  The catalytic subunit of yeast telomerase. Proc. Natl. Acad. Sci. USA 94:9202-9207[Abstract/Free Full Text].

CROUSE, G. F., 1996 Mismatch repair systems in Saccharomyces cerevisiae, pp. 411–448 in DNA Damage and Repair: Biochemistry, Genetics and Cell Biology, edited by J. NICKOLOFF and M. HOEKSTRA. Humana Press, Clifton, NJ.

CULLMANN, G., K. FIEN, R. B. KOBAYASHI, and B. STILLMAN, 1995  Characterization of the five replication factor C genes of Saccharomyces cerevisiae.. Mol. Cell. Biol. 15:4661-4671[Abstract].

DETLOFF, P., J. SIEBER, and T. D. PETES, 1991  Repair of specific base pair mismatches formed during meiotic recombination in the yeast Saccharomyces cerevisiae.. Mol. Cell. Biol. 11:737-745[Abstract/Free Full Text].

FEINSTEIN, S. I. and K. B. LOW, 1986  Hyper-recombining recipient strains in bacterial conjugation. Genetics 113:13-33[Abstract/Free Full Text].

FIEN, K. and B. STILLMAN, 1992  Identification of replication factor C from Saccharomyces cerevisiae: a component of the leading-strand DNA replication complex. Mol. Cell. Biol. 12:155-163[Abstract/Free Full Text].

GEITZ, R. D. and R. H. SCHIESTL, 1991  Applications of high efficiency lithium acetate transformation of intact yeast cells using single-stranded nucleic acids as carrier. Yeast 7:253-263[Medline].

GREEN, C. N. and S. JINKS-ROBERTSON, 1997  Frameshift intermediates in homopolymer runs are removed efficiently by yeast mismatch repair proteins. Mol. Cell. Biol. 17:2844-2850[Abstract].

HARRIS, R. S., K. J. ROSS, M. J. LOMBARDO, and S. M. ROSENBERG, 1998  Mismatch repair in Escherichia coli cells lacking single-strand exonucleases ExoI, ExoVII, and RecJ. J. Bacteriol. 180:989-993[Abstract/Free Full Text].

HAYNES, R. H., and B. A. KUNZ, 1981 DNA repair and mutagenesis, pp. 371–414 in The Molecular and Cellular Biology of the Yeast Saccharomyces: Life Cycle and Inheritance, edited by J. N. STRATHERN, E. W. JONES and J. R. BROACH. Cold Spring Harbor Laboratories, Cold Spring Harbor, NY.

HENDERSON, S. T. and T. D. PETES, 1992  Instability of simple sequence DNA in Saccharomyces cerevisiae.. Mol. Cell. Biol. 12:2749-2757[Abstract/Free Full Text].

HOLM, C., D. W. MEEKS-WAGNER, W. L. FANGMAN, and D. BOTSTEIN, 1986  A rapid, efficient method for isolating DNA from yeast. Gene 42:169-173[Medline].

HOWELL, E. A., M. A. MCALEAR, D. ROSE, and C. HOLM, 1994  CDC44: a putative nucleotide-binding protein required for cell cycle progression that has homology to subunits of replication factor C. Mol. Cell. Biol. 14:255-267[Abstract/Free Full Text].

JEYAPRAKASH, A., J. W. WELCH, and S. FOGEL, 1994  Mutagenesis of yeast MW104-1B strain has identified the uncharacterized PMS6 DNA mismatch repair gene locus and additional alleles of existing PMS1, PMS2 and MSH2 genes. Mutat. Res. 325:21-29[Medline].

JOHNSON, R. E., G. K. KOVVALI, L. PRAKASH, and S. PRAKASH, 1995  Requirement of the yeast RTH1 5' to 3' exonuclease for the stability of simple repetitive DNA. Science 269:238-240[Abstract/Free Full Text].