Spontaneous Frameshift Mutations in Saccharomyces cerevisiae: Accumulation During DNA Replication and Removal by Proofreading and Mismatch Repair Activities

Corresponding author: Sue Jinks-Robertson, Department of Biology, 1510 Clifton Rd., Emory University, Atlanta, GA 30322., jinks{at}biology.emory.edu (E-mail)

Communicating editor: M. LICHTEN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The accumulation of frameshift mutations during DNA synthesis is determined by the rate at which frameshift intermediates are generated during DNA polymerization and the efficiency with which frameshift intermediates are removed by DNA polymerase-associated exonucleolytic proofreading activity and/or the postreplicative mismatch repair machinery. To examine the relative contributions of these factors to replication fidelity in Saccharomyces cerevisiae, we determined the reversion rates and spectra of the lys2Bgl +1 frameshift allele. Wild-type and homozygous mutant diploid strains with all possible combinations of defects in the exonuclease activities of DNA polymerases and (conferred by the pol3-01 and pol2-4 alleles, respectively) and in mismatch repair (deletion of MSH2) were analyzed. Although there was no direct correlation between homopolymer run length and frameshift accumulation in the wild-type strain, such a correlation was evident in the triple mutant strain lacking all repair capacity. Furthermore, examination of strains defective in one or two repair activities revealed distinct biases in the removal of the corresponding frameshift intermediates by exonucleolytic proofreading and/or mismatch repair. Finally, these analyses suggest that the mismatch repair machinery may be important for generating some classes of frameshift mutations in yeast.

DNA replication is a highly accurate process with an overall in vivo error rate of less than one mutation per 10⁹ bases replicated per cell division (DRAKE et al. 1998 ). The fidelity of DNA synthesis in bacterial and eukaryotic cells is determined at three sequential levels (KUNKEL 1992 ; SCHAAPER 1993 ). First, the greatest contribution to replication fidelity is conferred by the inherent base selectivity of the DNA polymerase during nucleotide polymerization. Second, errors made by DNA polymerase can be corrected by a polymerase-associated exonucleolytic proofreading activity that removes terminal nucleotides that are incorrectly base paired with the template. Finally, a postreplicative mismatch repair (MMR) system removes replication errors that escape proofreading. Loss of the MMR system in human cells is associated with tumor formation (reviewed in BUERMEYER et al. 1999 ), thus underscoring the importance of efficiently removing DNA replication errors.

In the yeast Saccharomyces cerevisiae, three DNA polymerases (Pol , Pol , and Pol ) are important for the replication of genomic DNA (reviewed in SUGINO 1995 ). Pol has no associated 3' 5' exonuclease activity and appears to be involved only in primer synthesis. Pols and each have an associated 3' 5' exonucleolytic proofreading activity, and examination of mutation spectra in strains defective for either the exonuclease activity of Pol (pol3-01 mutants) or Pol (pol2-4 mutants) suggests that one polymerase is leading-strand specific and the other lagging-strand specific (MORRISON and SUGINO 1994 ; SHCHERBAKOVA and PAVLOV 1996 ; KARTHIKEYAN et al. 2000 ). It also has been demonstrated, however, that DNA replication can be completed in the absence of Pol catalytic activity, suggesting that Pol is at least capable of replicating both the leading and the lagging strands (KESTI et al. 1999 ). A combination of the pol2-4 and pol3-01 alleles is synthetically lethal in haploids and results in a synergistic increase in mutation rate (MORRISON and SUGINO 1994 ), suggesting that the exonuclease activities of Pols and are partially redundant and compete for a common substrate(s). In addition to their roles in proofreading, the exonuclease activities of Pols and may be involved in MMR, where their associated 3' 5' exonuclease activities have been proposed to act in concert with the 5' 3' exonuclease activity of Exo1p to remove mismatches (TRAN et al. 1999 ). Finally, recent work has demonstrated that presence of the pol3-01 allele is associated with checkpoint-dependent delays in entering and transiting S phase (DATTA et al. 2000 ). Surprisingly, the pol3-01-associated mutator phenotype is partially dependent on the S-phase checkpoint, because mutation rates decrease in checkpoint-defective mutants (DATTA et al. 2000 ).

The postreplicative MMR system is responsible for correcting polymerization errors that escape proofreading. The best-understood MMR system is the methyl-directed MutHLS system of Escherichia coli, where MutL couples MutS mismatch recognition to the downstream processing steps (reviewed in MODRICH and LAHUE 1996 ). In yeast, there are six MutS homologs (Msh1p–Msh6p) and four MutL homologs (Pms1p and Mlh1–Mlh3p; reviewed in HARFE and JINKS-ROBERTSON 2000 ). Msh2p is required for all nuclear mismatch repair, and mismatch recognition is effected by heterodimers of Msh2p with either Msh3p or Msh6p (JOHNSON et al. 1996 ; MARSISCHKY et al. 1996 ). Although Mlh1p forms heterodimers with Pms1p, Mlh2p, and Mlh3p (WANG et al. 1999 ), most MMR involves the Mlh1p/Pms1p heterodimer. A combination of proofreading-defective Pol (pol3-01 allele) with either a pms1 or msh2 allele results in synthetic lethality in haploids, but homozygous diploid strains are viable (MORRISON et al. 1993 ; TRAN et al. 1999 ). In contrast to the situation with the pol3-01 allele, haploid strains with a proofreading-defective Pol (pol2-4 allele) and either a pms1 or msh2 allele have been isolated (MORRISON et al. 1993 ; TRAN et al. 1997 , TRAN et al. 1999 ). Mutation-rate measurements in double-mutant haploid and diploid strains are consistent with the notion that the yeast proofreading and MMR activities act in a series to correct replication errors (MORRISON et al. 1993 ).

Replication errors typically can be classified as base substitution events or as insertion/deletion events involving a small number of nucleotides. An insertion/deletion that is not a multiple of 3 bp alters the reading frame of the corresponding gene and almost always eliminates gene function. Because of the highly deleterious nature of frameshift mutations, it is important to understand the mechanisms for generating insertions/deletions during DNA replication as well as the editing functions that prevent fixation of such replication errors. The most widely recognized model of frameshift mutagenesis is the "direct slippage" model, in which DNA polymerase slippage within a tandemly repeated sequence leads to the deletion or addition of one or more repeat units (STREISINGER et al. 1966 ). This model is supported by numerous in vivo studies, which have demonstrated that tandemly repeated sequences such as mono- and dinucleotide repeats are hotspots for frameshift mutations (see JINKS-ROBERTSON et al. 1998 for a review of yeast studies). The other two models of frameshift mutagenesis are based strictly on in vitro studies. In the "misincorporation-realignment" model, a misincorporation by DNA polymerase initiates the slippage event, which restores correct base pairing between the 3' end of the nascent strand and the template (KUNKEL and SONI 1988 ). In the "dNTP-mediated misalignment" model, the misalignment occurs before, rather than after, the incorporation of an incorrect nucleotide, which is specified by the next template base and therefore is not random (BLOOM et al. 1997 ). Because the latter two models do not necessarily involve tandemly repeated sequences, they may account for in vivo frameshift mutations that involve noniterated sequences.

We and others have used the yeast lys2Bgl frameshift reversion assay to obtain frameshift mutation rates and spectra in wild-type cells, in cells defective in various MMR components (MARSISCHKY et al. 1996 ; GREENE and JINKS-ROBERTSON 1997 ; FLORES-ROZAS and KOLODNER 1998 ) or in cells defective in the proofreading activity of Pol (DATTA et al. 2000 ). In the current study, these analyses are extended by examining lys2Bgl reversion rates and spectra in wild-type and in completely MMR-defective strains that are deficient in the individual proofreading activities of Pols and or that are simultaneously deficient in both proofreading activities. These analyses reveal the most frequent frameshift errors made during nucleotide polymerization and provide novel insight into the relative contributions of individual proofreading activities and MMR to the overall stability of the yeast genome.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media and growth conditions:
Yeast strains were grown in standard media (SHERMAN 1991 ) at 30°. Cells were grown nonselectively in YEP (1% yeast extract, 2% Bacto peptone, and 2.5% agar for plates) supplemented with 2% glycerol/2% ethanol (YEPGE) or 2% dextrose (YPD). Selective growth was in synthetic complete (SC) medium containing 2% dextrose and lacking the appropriate amino acid. Ura^- yeast segregants were selected on medium containing 5-fluoroorotic acid (5-FOA; BOEKE et al. 1987 ).

Strain constructions:
All diploid strains used in this study were derived by mating isogenic derivatives of haploid strains SJR335 and SJR357. A complete list of the haploid strains is given in Table 1. Mutant alleles were introduced into SJR335 and SJR357 by standard transformation procedures (GIETZ et al. 1992 ). The msh2 allele was introduced using AatII/XbaI-digested GC1914 (GREENE and JINKS-ROBERTSON 1997 ) and confirmed by PCR. The pol2-4 and pol3-01 alleles were introduced by two-step gene replacement using the URA3-containing plasmids YIpJB1 (MORRISON et al. 1991 ) and YIpAM26 (MORRISON et al. 1993 ), respectively. Presence of the pol2-4 allele was confirmed by allele-specific PCR (MORRISON and SUGINO 1994 ); presence of the pol3-01 allele was confirmed using an associated restriction site polymorphism (MORRISON et al. 1993 ). The leu2-R allele was introduced into SJR357 derivatives by two-step gene replacement using pJH189 (LICHTEN et al. 1987 ). Because combinations of most repair-deficient mutant alleles are lethal in haploids, one mutant allele was introduced, followed by a complementing plasmid and then the second mutant allele. Haploids were then mated and diploids that had lost the complementing plasmid were identified. Two independent isolates of each diploid strain, which were derived by mating independent haploid isolates, were used for measurements of reversion rates and determination of mutation spectra.

Strains SJR823 and SJR824 were mated to create the msh2/msh2 pol2-4/pol2-4 diploid. To construct SJR823, the msh2 strain SJR480 was first transformed with pSR578 (MSH2-HIS3-CEN plasmid; our laboratory collection) and then the pol2-4 allele was introduced to create SJR823. SJR824 was constructed using the same approach, but starting with msh2 strain SJR685. SJR824 was transformed with plasmid GC1913 (MSH2-URA3-CEN plasmid; obtained from G. F. Crouse) and then was mated to SJR823; the resulting diploids were grown nonselectively to allow loss of pSR578 (His^- segregants). Following loss of pSR578, diploids were plated on 5-FOA to selectively identify loss of plasmid GC1913.

Strains SJR918 and SJR919 were mated to create the msh2/msh2 pol3-01/pol3-01 diploid. To construct SJR918, the pol3-01 and leu2-R alleles were introduced into SJR357, creating SJR882. SJR882 was transformed with the POL3-containing plasmid HL1 (POL3-LEU2-CEN; GORDENIN et al. 1992 ), creating SJR890, and then the msh2 allele was introduced into SJR890 to give SJR918. SJR919 was constructed by introducing pol3-01 into a msh2 strain (SJR685) containing an MSH2-complementing plasmid (pSR578). SJR919 was transformed with GC1913 (MSH2-URA3-CEN plasmid) prior to mating with SJR918. The resulting diploid was grown nonselectively to allow loss of HL1 and pSR578 (Leu^- and His^- segregants, respectively) and dilutions were then plated on 5-FOA to select loss of plasmid GC1913.

Strains SJR920 and SJR921 were mated to create the pol2-4/pol2-4 pol3-01/pol3-01 diploid. SJR920 was constructed by introducing the pol2-4 allele into SJR890, a pol3-01 strain containing the POL3-complementing plasmid HLI. SJR921 was similarly constructed starting with the pol3-01 strain SJR722 containing plasmid HL1. SJR921 was transformed with pBL304 (POL3-URA3-CEN; GORDENIN et al. 1992 ) and mated with SJR920. Diploid cells were grown nonselectively to allow loss of HL1 (Leu^- segregants) and dilutions were then plated on 5-FOA to select for loss of the plasmid pBL304.

Strains SJR1179 and SJR1180 were mated to create the pol2-4/pol2-4 pol3-01/pol3-01 msh2/msh2 triple-mutant diploid. SJR1179 and SJR1180 were constructed by introducing the pol2-4 allele into pol3-01 msh2 haploid strains (SJR918 and SJR919, respectively), which had previously been transformed with MSH2- and POL3-complementing plasmids (pSR578 and HL1, respectively). SJR1179 and SJR1180 were mated, and diploids were transformed with the plasmid pBL304 (POL3-URA3-CEN). The diploids were grown nonselectively to allow loss of the plasmids pSR578 and HL1 (His^- and Leu^- segregants, respectively), and dilutions were then plated on 5-FOA to select for loss of plasmid pBL304.

Reversion rates and spectra:
For rate determinations, 2-day-old colonies were taken from YEPD plates, inoculated into 5 ml YEPGE liquid medium, and grown for 2 days on a roller drum. Cells were harvested by centrifugation, washed once with sterile H₂O, and resuspended in 1 ml of H₂O. Aliquots (100 µl) of appropriate dilutions were plated on SC-Lys to select Lys⁺ revertants and on YEPD to determine viable cell numbers. Lys⁺ colonies were counted on day 3 after selective plating. Because of slow growth, the pol2-4 pol3-01 msh2 triple mutant was grown 3 days in YEPGE and Lys⁺ revertants were counted on day 5 after selective plating. Reversion rates were determined by the method of the median (LEA and COULSON 1949 ), using data from 10 to 20 cultures of each strain. The 95% confidence intervals for the rates were calculated as described by DIXON and MASSEY 1969 .

To isolate independent Lys⁺ revertants for DNA sequence analysis, 1-ml YEPGE cultures were grown as described above and a single aliquot was plated on SC-Lys. One revertant from each culture was purified for subsequent molecular analysis. Standard dideoxy DNA sequencing of revertants was performed as described by GREENE and JINKS-ROBERTSON (1997). Pairwise comparisons of mutation spectra were done using an algorithm developed by Adams and Skopek (see CARIELLO et al. 1994 ). All spectral comparisons yielded highly significant P values.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The lys2Bgl allele is the result of a GATC insertion into the BglII site in the N-terminal portion of LYS2 (GREENE and JINKS-ROBERTSON 1997 ), and reversion of this allele was used to assess frameshift mutagenesis. The lys2Bgl allele reverts by compensatory 3N - 1 frameshift events, which are constrained by stop codons in alternative reading frames to occur within an 150-bp "reversion window" surrounding the lys2Bgl mutation. The reversion window contains several mononucleotide runs as well as extensive stretches of nonrepetitive sequence, thus allowing the identification of a wide variety of frameshift mutations. To assess the relative roles of proofreading and MMR in replication fidelity, the yeast MMR machinery was inactivated by deletion of MSH2 and the proofreading activities of Pols and were inactivated using appropriate exonuclease-deficient alleles (pol3-01 and pol2-4, respectively). Diploid strains were used in all experiments because of the documented synthetic lethality between pol3-01 and pms1 or msh2 alleles (MORRISON et al. 1993 ; TRAN et al. 1999 ) and between pol2-4 and pol3-01 alleles (MORRISON and SUGINO 1994 ) in haploid yeast strains. Although disruption of MSH2 in a pol2-4 haploid strain has been reported (TRAN et al. 1999 ), our repeated attempts to similarly disrupt MSH2 in our pol2-4 strains were unsuccessful. In addition, pol2-4 haploid strains containing a MSH2-complementing plasmid were unable to lose the plasmid after disruption of MSH2. A combination of msh2 with pol2-4 thus is synthetically lethal in our haploid strain backgrounds.

lys2Bgl reversion rates:
The reversion rates of the lys2Bgl allele in wild-type and various single-, double-, and triple-mutant strains are given in Table 2. The increases in reversion rate of lys2Bgl in the pol2-4 mutant (a 13-fold increase), the pol3-01 mutant (a 300-fold increase), and the msh2 mutant (a 200-fold increase) are consistent with previously reported reversion rate increases for the his7-2 frameshift allele (MORRISON and SUGINO 1994 ). The reversion rate increases in the pol2-4 msh2 and pol3-01 msh2 strains relative to the single mutant strains are approximately multiplicative (2300-fold and 16,000-fold, respectively), a behavior that is consistent with exonucleolytic proofreading and MMR acting sequentially on the same frameshift intermediates (MORRISON et al. 1993 ). Combination of pol2-4 and pol3-01 results in a synergistic, 11,000-fold increase in the lys2Bgl reversion rate, which is in agreement with similar measurements of his7-2 reversion rate (MORRISON and SUGINO 1994 ). Surprisingly, the pol2-4 pol3-01 msh2 triple mutant exhibits no significant change in the lys2Bgl reversion rate relative to the pol2-4 pol3-01 double mutant. As has been argued for E. coli (FIJALKOWSKA and SCHAAPER 1996 ), the lack of an increase in mutation rate in the triple mutant could simply be due to a saturation of the mismatch repair system in the pol2-1 pol3-01 double mutant. Examination of the corresponding mutation spectra, however, suggests that this explanation does not adequately account for the rate results in yeast (see DISCUSSION). It should be noted that a pol2-4 pol3-01 msh2 triple-mutant strain defective in all repair capacity has not been previously described. In addition to having a highly elevated mutation rate, the triple-mutant strain grew relatively slowly in liquid culture, had a low plating efficiency (only 50% of cells produced colonies), and produced colonies of variable size when plated (data not shown).

View this table: In this window In a new window   Table 2. Rates and distributions of lys2Bgl reversion events

Reversion spectra in wild-type and triple mutant strains:
The relative efficiencies of the proofreading and MMR systems in removing different types of frameshift intermediates can be inferred by comparing the reversion spectra derived from mutant vs. wild-type strains. If a given repair system removes all intermediates with the same efficiency, then the mutant spectrum should resemble the wild-type spectrum. If, however, some frameshift intermediates are removed with greater efficiency than are other intermediates, then the spectra of wild-type vs. mutant strains will be different. Specifically, those frameshift intermediates that are corrected most efficiently by a given repair system will comprise a greater proportion of the mutation spectrum in the mutant than in the wild-type strain.

The reversion events observed in the wild-type diploid (Fig 1A) presumably reflect polymerization errors that are corrected neither by proofreading nor by the postreplicative MMR system, and the spectrum is very similar to that reported previously with one of the haploid parental strains (GREENE and JINKS-ROBERTSON 1997 ). Single base deletions account for 95% (97/102) of the reversion events in the diploid; the remaining events are composed of four 2-bp insertions and a single complex event in which a base substitution accompanies a 1-bp deletion. In an earlier analysis of lys2Bgl reversion, homopolymer runs >3N were hotspots for frameshift events, as they accumulated frameshift events more often than would be predicted for noniterated sequences of the same length (GREENE and JINKS-ROBERTSON 1997 ). In the current study, 58% (59/102) of the 1-bp deletions in the wild-type diploid strain occur in the four monotonic runs that are >3N: the 6A, 5T, 4C, and 4A runs beginning at nucleotides (nt) 664, 720, 697, and 727, respectively, and accounting for 13% of the reversion window. The 6A and 4C runs are particularly prone to accumulating -1 frameshifts and account for 49% (50/102) of the total reversion events.

View larger version (29K): In this window In a new window Download PPT slide   Figure 1. Frameshift reversion spectra for a wild-type strain and for strains defective in proofreading and/or mismatch repair. The +4 insertion that creates the lys2Bgl allele is underlined and in boldface type. Single base deletions are represented by individual 's below the sequence and insertions are indicated above the sequence. N is the total number of revertants sequenced to obtain each spectrum. Pairwise comparisons of the wild-type (WT) spectrum with each mutant spectrum yielded highly significant P values (P < 0.01), with the exception of the WT-msh2 comparison (P = 0.076). This exception is likely related to the relatively small sample size of the msh2 Lys+ revertants sequenced (N = 45), as pooling the diploid data reported here with the haploid data reported previously (GREENE and JINKS-ROBERTSON 1997 ) yielded a highly significant P value.

Examination of the mutation spectrum in a strain simultaneously defective for MMR and for the proofreading activities of both Pol and Pol should, in principle, reflect the frameshift errors that occur during DNA replication. The mutation spectrum from the pol2-4 pol3-01 msh2 triple mutant demonstrates that frameshift mutations accumulate primarily in homopolymer runs, with 83% (81/98) of events occurring in runs >3N (Fig 1B). In contrast to the spectrum obtained from the wild-type background, however, the frequency of events in the runs is directly proportional to the run length. The 6A run thus accounts for 50% (49/98; 48 1-nt deletions, one 2-nt insertion) of the total events, followed by 19% (19/98) of the events in the 5T run and 7% of the events in each of the 4N runs (6/98 and 8/98 in the 4C and 4A runs, respectively). Although 3N runs are not considered hotspots for frameshift events in this system, the 3T run at position 713 accounts for 7% (7/98) of the events in the triple mutant. There are nine 3N runs in the reversion window, and 7 of the 11 frameshift events that occur in the 3N runs are in the nucleotide 713 3T hotspot. This particular run was previously shown to be unique among the 3N runs, as it was found to be a novel deletion hotspot in a msh6 strain (GREENE and JINKS-ROBERTSON 1997 ). We speculate that the 3T run at nucleotide 713 may be particularly prone to frameshift events because of sequence context.

Reversion spectra in completely proofreading-defective (pol2-4 pol3-01 double mutant) or MMR-defective (msh2) strains:
In contrast to frameshift events in the wild-type and triple mutant strains, the frameshift events in a pol2-4 pol3-01 double-mutant strain do not occur preferentially in homopolymer runs >3N (Fig 1C). Sixty-three percent (57/91) of the 1-bp deletions occur in noniterated sequences and three prominent deletion hotspots account for 51% (46/91) of the total events: G676, C773, and C778. Comparison of these three sites yields the consensus sequence of 5'-CTTTG-3', with deletion of the cytosine comprising the selected frameshift event. Two other hotspots are at GG dinucleotide repeats at positions 689 and 770, with each accounting for 6% of the total events (7/91 and 5/91, respectively).

Relative to the wild-type and triple-mutant strains, the reversion spectrum for the msh2 diploid strain shows a dramatic increase in the proportion of events in the 6A and 4C runs (Fig 1D), which is similar to previous results obtained in haploid strain backgrounds (MARSISCHKY et al. 1996 ; GREENE and JINKS-ROBERTSON 1997 ). The 6A and 4C runs account for 96% (43/45) of the frameshift events, compared to 49% of the events in the wild-type strain and 55% of the events in the triple-mutant strain.

Reversion spectra in strains defective in the exonucleolytic proofreading activity of either Pol (pol3-01) or Pol (pol2-4):
In addition to obtaining frameshift spectra in either the presence or the absence of the exonuclease activities of both Pol and Pol , we also analyzed strains defective in the exonuclease activity of only a single polymerase (Fig 2). This analysis was done in both MMR-proficient and MMR-deficient backgrounds. The spectra obtained in the absence of the Pol vs. the Pol exonuclease activity are strikingly different. Elimination of only the Pol exonuclease activity (pol2-4 allele; Fig 2A and Fig B) results in a clustering of the frameshift events at the 6A and 4C homopolymer runs, which is reminiscent of the pattern obtained in the MMR-deficient background (Fig 1D). In the pol2-4 single-mutant spectrum, 86% (83/97) of the 1-bp deletions are the 6A or 4C run, with a 3:1 bias for events in the 4C run (Fig 2A). In the pol2-4 msh2 double mutant, however, almost all -1 events (68/80 = 85%) are in the 6A run and very few events (2/80 = 3%) are in the 4C run (Fig 2B).

View larger version (32K): In this window In a new window Download PPT slide   Figure 2. Frameshift reversion spectra for strains deficient in the exonuclease activities of individual polymerases. See Fig 1 for further explanation.

In contrast to the clustering of frameshift events in homopolymer runs in the pol2-4 mutant, events in the pol3-01 mutant are more variable. In the pol3-01 single mutant, only 33% (31/95) of the frameshifts are in the 6A or 4C run, and events are distributed evenly between the two runs (Fig 2C). Twenty-five percent (24/95) of the 1-bp deletions involve noniterated sequences and the hotspot at G676 that was seen in the pol2-4 pol3-01 double mutant is evident, corresponding to 75% (18/24) of the events in noniterated sequences. Interestingly, the other two 1N deletion hotspots evident in the pol2-4 pol3-01 double mutant (Fig 1C) are not distinct hotspots when the exonuclease activity of only one polymerase is defective. In the pol3-01 msh2 double mutant, 62% (58/93) of the events are in the runs >3N, and there is a 4:1 bias for events in the 4C run vs. the 6A run.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

One means of assessing the relative roles of proofreading and MMR in removing mutational intermediates is to compare mutation rates and spectra in wild-type or completely repair-defective cells to those obtained in cells defective for either proofreading or MMR. Such an approach has been successfully applied in E. coli using a forward mutation system that detects primarily base substitutions (SCHAAPER 1993 ), and we have undertaken a similar strategy in yeast using the frameshift-specific lys2Bgl reversion assay. To facilitate comparisons of relevant mutation spectra shown in Fig 1, the distributions of lys2Bgl reversion events are graphically summarized in Fig 3.

View larger version (22K): In this window In a new window Download PPT slide   Figure 3. Proportion of deletion events in noniterated sequences and in homopolymer runs of various length (N). Because of the difference in the behaviors of the 4C and 4A runs, these are treated separately. (A) Wild-type strain. (B) pol2-4 pol3-01 msh2 strain. (C) pol2-4 pol3-01 MSH2 strain. (D) POL2 POL3 msh2 strain.

Proofreading by DNA polymerases provides the first step for editing potential frameshift intermediates. In principle, eliminating the exonuclease activity of either Pol or Pol should reveal the proofreading specificity of the corresponding polymerase and reflect the underlying polymerization errors (see MORRISON and SUGINO 1994 for a discussion). In our frameshift-specific assay, as in other assays (MORRISON and SUGINO 1994 ; SHCHERBAKOVA and PAVLOV 1996 ; KARTHIKEYAN et al. 2000 ), the mutation spectra generated in the pol2-4 vs. pol3-01 single mutants were strikingly different, which is consistent with distinct roles of the polymerases during DNA replication. A synergistic increase in mutation rate similar to that observed for the lys2Bgl allele has been previously documented in pol2-4 pol3-01 double mutants (MORRISON and SUGINO 1994 ), but no corresponding mutation spectra have been reported. In the frameshift-specific assay used here, the rates of frameshift mutations increased in both noniterated (1N) sequences and mononucleotide runs (2N–6N) in the double mutant relative to the wild-type strain, with the largest increase being observed in 1N sequences and the smallest increase in the 6N sequence (60,000-fold and 1700-fold, respectively; Table 2 and Fig 3). The shift in the distribution of 1-bp deletions in the pol2-4 pol3-01 double mutant is consistent with the notion that nonrepetitive sequences are more efficient substrates for proofreading than are repetitive sequences (KUNKEL and BEBENEK 2000 ). Interestingly, there were two notable 1N hotspots (C773 and C778) in the pol2-4 pol3-01 double-mutant spectrum that were not evident in either single-mutant spectrum. The presence of these hotspots only in the double mutant is consistent with the functional redundancy between the Pol and Pol exonuclease activities as deduced from mutation rate measurements. It has been suggested that the synergism may reflect either the ability of one polymerase to proofread the mistakes of the other (MORRISON and SUGINO 1994 ) or roles of the exonuclease activities in MMR processes as well as in proofreading (TRAN et al. 1999 ). It should be noted that the spectrum of mutations in the pol2-4 pol3-01 double mutant is distinctly different from that of an MMR-defective (msh2) strain, an observation that does not support a concomitant defect in MMR.

Although useful information concerning polymerase fidelity can be obtained by comparing exonuclease-proficient and exonuclease-deficient strains that are otherwise wild type, an alternative way to assess the role of proofreading in mutation avoidance is to make the comparison in strains that are MMR defective (i.e., msh2 vs. pol2-4 po3-01 msh2). This latter type of comparison not only eliminates the replication-editing function of the MMR machinery, which can greatly impact mutation rates and spectra, but also eliminates any potential contributions that polymerase-associated exonuclease activities might have to MMR. The lys2Bgl reversion rate was elevated 25-fold in the pol2-4 po3-01 msh2 strain relative to the msh2 strain (Table 2) and there was a notable shift in the distribution of frameshift events within runs >3N (compare Fig 3B and Fig D). Specifically, the data suggest that proofreading is more efficient in the 4A and 5T runs than in the 4C and 6A runs. The relatively inefficient proofreading of slippage events in the 6A run is most likely due to the length of the run, as it is near the proofreading threshold deduced from in vitro and in vivo experiments (KROUTIL et al. 1996 ; TRAN et al. 1997 ). We suggest that the relatively inefficient removal of frameshift intermediates in the 4C run is related to sequence composition, although sequence context also may be important. Studies using bacteriophage T4 indeed have implicated base composition as an important determinant of proofreading efficiency, and it has been suggested that stable (i.e., GC-rich) regions of DNA are less likely to have hydrogen bonding disrupted and should, therefore, be less available to exonucleolytic proofreading activity (BESSMAN and REHA-KRANTZ 1977 ; GOODMAN and FYGENSON 1998 ).

The MMR pathway represents the last step for editing DNA polymerization, and its contribution to mutation avoidance was estimated by comparing the lys2Bgl reversion rates and spectra in wild-type vs. msh2 strains. Loss of MMR activity was accompanied by a 200-fold increase in reversion rate, and there was a striking shift in the distribution of frameshift mutations (Table 2 and Fig 3). One interpretation of the run-associated clustering of mutations upon loss of Msh2p is that the yeast MMR system removes frameshift intermediates in runs >3N more efficiently than those in shorter runs or noniterated sequence. It is difficult, however, to imagine a mechanism whereby the MMR system might recognize and repair an extrahelical base in a run better than that in a noniterated sequence. It seems more likely that the frameshifts that are underrepresented in the msh2 spectrum might arise out of the context of normal DNA replication (e.g., during DNA repair or bypass), where they either may not be sensed by the MMR machinery or may not be subject to the strand bias normally associated with MMR. Alternatively, it is formally possible that a functional MMR system is required to generate some classes of frameshift mutations (see below).

In addition to comparing wild-type and MMR-defective strains, we also examined MMR specificity under conditions where proofreading was not contributing to error avoidance. Our expectation was that the pol2-4 pol3-01 msh2 triple mutant would exhibit a greatly elevated mutation rate relative to the pol2-4 pol3-01 double mutant, but surprisingly, the lys2Bgl reversion rates were not statistically different in the two strains. Although the simplest explanation for the lack of a further increase in mutation rate in the triple mutant is that the MMR system was already saturated in the pol2-4 pol3-01 double mutant (see FIJALKOWSKA and SCHAAPER 1996 ), this explanation cannot account for the dramatic shift in the mutation spectrum observed in the triple mutant. Almost 60% (57/91) of the frameshift events in the double mutant were in 1N sequences, for example, and yet there were no events in 1N sequences among 98 revertants sequenced from the triple mutant. In addition, there was a sevenfold decrease in the rate of events in 2N runs in the triple mutant relative to the double mutant. This spectral shift suggests that a functional MMR system may be required to generate specific classes of frameshift intermediates, most notably those in nonrun (1N and 2N) sequences (see below for further discussion).

The majority of frameshift events are assumed to arise in runs of repeated sequence, where the potential number of correct base pairs that can stabilize a slippage-generated frameshift intermediate, as well as the total number of potential intermediates, increases as the run length increases (KUNKEL and BEBENEK 2000 ). This leads to the prediction that the slippage rate should increase as the length of the run increases and this prediction has been confirmed in in vitro DNA replication assays (KROUTIL et al. 1996 ) and in vivo yeast studies (TRAN et al. 1997 ). As expected, the majority (58%) of deletion events in the wild-type spectrum occurred in mononucleotide runs longer than 3N, which together comprise only 13% of the lys2Bgl reversion window. The distribution of events, however, was not consistent with run length being the primary determinant of frameshift accumulation (Fig 3A and GREENE and JINKS-ROBERTSON 1997 ). Specifically, the proportions of deletions in the two 4N runs were not equal (5/102 and 22/102 events in the 4A and 4C runs, respectively), and the 5T run was no more likely to accumulate frameshift mutations than random sequence of the same length.

Because the pol2-4 pol3-01 msh2 triple-mutant strain lacks both proofreading and MMR, the corresponding frameshift spectrum should provide an accurate reflection of the errors made during replicative DNA synthesis. Not only was the mutation rate elevated 5200-fold in the triple mutant relative to the wild-type strain (Table 2), but there also was a shift in the mutation spectrum. There was a larger proportion of deletion events in mononucleotide runs >3N in the triple mutant than in the wild-type strain (83% vs. 58%) and the distributions of events between the runs >3N differed significantly in two strains (P < 0.01 by contingency chi-square). Most notably the distribution of 1-bp deletion events was correlated with increasing mononucleotide run length in the triple mutant, with the 6N run accounting for the majority of events, followed by the 5N run and then the two 4N runs (Fig 3B). As there was a direct correlation between run length and frameshift distribution in the triple mutant, the apparent run specificity for deletions observed in the wild-type strain can be attributed to differential repair of frameshift intermediates rather than to preferential polymerase errors.

As discussed above, the data obtained with the wild-type, pol2-4 pol3-01 double-mutant, and pol2-4 pol3-01 msh2 triple-mutant strains fit the general predictions that slippage frequency should be directly proportional and proofreading efficiency should be inversely proportional to run length (KUNKEL and BEBENEK 2000 ). Two implicit assumptions were made, however, in interpreting the data. First, we assumed that the elimination of polymerase-associated exonuclease activity affects only the proofreading of DNA replication errors, and second, we assumed that elimination of Msh2p impacts only postreplicative MMR. If the first assumption is indeed true, then the frameshift mutations observed in the pol2-4 pol3-01 strain should directly reflect the primary slippage errors generated by the corresponding wild-type polymerases. It is possible, however, that the nature of the primary polymerization errors might be altered in the absence of proofreading capacity. In the absence of proofreading, for example, a base substitution intermediate might have a much higher probability of being converted into a frameshift intermediate by slippage between the nascent and template strands (see BEBENEK and KUNKEL 1990 ). Such slippage would restore base pairing between the 3' end of the nascent strand and the template, thus allowing a polymerase that cannot go backward to proceed forward. Mispair-promoted slippage could occur either during normal replicative DNA synthesis or in association with a checkpoint response to a mispaired 3' end, in which case it might involve a translesion polymerase (see DATTA et al. 2000 ). It follows that the types and proportions of frameshift intermediates generated by a proofreading-defective polymerase might be different from those generated by a more processive, proofreading-proficient polymerase.

The second assumption inherent in our analyses is that the only mutation-related process affected by removal of Msh2p is postreplicative MMR. In mammalian cells, however, there is evidence that MSH2 also is involved in triggering apoptosis in response to DNA-damaging agents, and it has been suggested that the MSH2 may function as a general damage sensor (GONG et al. 1999 ; TOFT et al. 1999 ). Although yeast cells do not undergo apoptosis in response to DNA damage, they do delay cell-cycle progression (FOIANI et al. 2000 ). One intriguing possibility is that yeast Msh2p is involved in recognizing DNA damage (or perhaps aberrant replication intermediates) and may thereby be important in triggering a checkpoint response that leads to novel types of frame-shift intermediates. Such a scenario could account for the observation that elimination of Msh2p in either a wild-type or a pol2-4 pol3-01 background was accompanied by a proportional decrease of frameshift events in nonrun sequences.

In summary, the data reported here provide a comprehensive analysis of frameshift mutagenesis in yeast strains that are singly, doubly, or triply defective in the MMR and the proofreading activities of Pols and . These analyses indicate very different specificities for Pol and Pol in the generation and/or removal of frameshift intermediates, even though loss of the exonuclease activity of one polymerase can be partially compensated for by that of the other polymerase. In addition, comparisons of frameshift spectra in MMR-proficient and MMR-deficient strains suggest either that the efficiency with which the MMR system removes extrahelical bases is greatly influenced by sequence context or that the generation of some classes of frameshifts is actually dependent on the presence of the MMR system. These results affirm the complexities and the highly interconnected natures of the pathways that generate and remove -1 frameshift intermediates, and one can expect similar complexities to emerge in analyses of other types of mutational intermediates.

	FOOTNOTES

¹ Present address: Centers for Disease Control and Prevention, National Center for Environmental Health, Atlanta, GA 33041.

	ACKNOWLEDGMENTS

We thank Miyono Hendrix for technical assistance and A. Datta and members of the SJR lab for critical comments on the manuscript. We also thank G. F. Crouse, D. Gordenin, and A. Sugino for providing plasmids used in the strain constructions. This work was supported by a grant from the National Science Foundation to S.J.-R. C.N.G. was supported in part by the Graduate Division of Biological and Biomedical Sciences at Emory University.

Manuscript received March 12, 2001; Accepted for publication June 6, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BEBENEK, K. and T. A. KUNKEL, 1990  Frameshift errors initiated by nucleotide misincorporation. Proc. Natl. Acad. Sci. USA 87:4946-4950[Abstract/Free Full Text].

BESSMAN, M. J. and J. REHA-KRANTZ, 1977  Studies on the biochemical basis of spontaneous mutation. J. Mol. Biol. 116:115-123[Medline].

BLOOM, L. B., X. CHEN, D. K. FYGENSON, J. TURNER, and N. O'DONNELL et al., 1997  Fidelity of Escherichia coli DNA polymerase III holoenzyme. J. Biol. Chem. 272:27919-27930[Abstract/Free Full Text].

BOEKE, J. D., J. TRUEHEART, G. NATSOULIS, and G. R. FINK, 1987  5-Fluoroorotic acid as a selective agent in yeast molecular genetics. Methods Enzymol. 154:164-175[Medline].

BUERMEYER, A., S. M. DESCHENES, S. M. BAKER, and R. M. LISKAY, 1999  Mammalian DNA mismatch repair. Annu. Rev. Genet. 33:533-564[Medline].

CARIELLO, N. F., W. W. PIEGORSCH, W. T. ADAMS, and T. R. SKOPEK, 1994  Computer program for the analysis of mutational spectra: application to p53 mutations. Carcinogenesis 15:2281-2285[Abstract/Free Full Text].

DATTA, A., J. L. SCHMEITS, N. S. AMIN, P. J. LAU, and K. MYUNG et al., 2000  Checkpoint dependent activation of mutagenic repair pathways in Saccharomyces cerevisiae pol3–01 mutants. Mol. Cell 6:593-603[Medline].

DIXON, W. J., and F. J. MASSEY, JR., 1969 Introduction to Statistical Analysis. McGraw-Hill, New York.

DRAKE, J. W., B. CHARLESWORTH, D. CHARLESWORTH, and J. F. CROW, 1998  Rates of spontaneous mutation. Genetics 148:1667-1686[Abstract/Free Full Text].

FIJALKOWSKA, I. J. and R. M. SCHAAPER, 1996  Mutants in the Exo I motif of Escherichia coli dnaQ: defective proofreading and inviability due to error catastrophe. Proc. Natl. Acad. Sci. USA 93:2856-2861[Abstract/Free Full Text].

FLORES-ROZAS, H. and R. D. KOLODNER, 1998  The Saccharomyces cerevisiae MLH3 gene functions in MSH3-dependent suppression of frameshift mutations. Proc. Natl. Acad. Sci. USA 95:12404-12409[Abstract/Free Full Text].

FOIANI, M., A. PELLICIOLI, M. LOPES, C. LUCCA, and M. FERRARI et al., 2000  DNA damage checkpoints and DNA replication controls in Saccharomyces cerevisiae. Mutat. Res. 451:187-196[Medline].

GIETZ, D., A. ST. JEAN, R. A. WOODS, and R. H. SCHIESTL, 1992  Improved method for high efficiency transformation of intact yeast cells. Nucleic Acids Res. 20:1425[Free Full Text].

GONG, J., A. CONSTANZO, H.-Q. YANG, G. MELINO, and W. G. KAELIN, JR. et al., 1999  The tyrosine kinase c-Abl regulates p73 in apoptotic response to cisplatin-induced DNA damage. Nature 399:806-809[Medline].

GOODMAN, M. F. and D. K. FYGENSON, 1998  DNA polymerase fidelity: from genetics toward a biochemical understanding. Genetics 148:1475-1482[Abstract/Free Full Text].

GORDENIN, D. A., A. L. MALKOVA, A. PETERZEN, V. N. KULIKOV, and Y. I. PAVLOV et al., 1992  Transposon Tn5 excision in yeast: influence of DNA polymerases , , and and repair genes. Proc. Natl. Acad. Sci. USA 89:3785-3789[Abstract/Free Full Text].

GREENE, 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].

HARFE, B. D. and S. JINKS-ROBERTSON, 2000  DNA mismatch repair and genetic instability. Annu. Rev. Genet. 34:359-399[Medline].

JINKS-ROBERTSON, S., C. GREENE and W. CHEN, 1998 Genetic instabilities in yeast, pp. 485–507 in Genetic Instabilities and Hereditary Neurological Diseases, edited by R. D. WELLS and S. T. WARREN. Academic Press, San Diego.

JOHNSON, R. E., G. K. KOVVALI, L. PRAKASH, and S. PRAKASH, 1996  Requirement of the yeast MSH3 and MSH6 genes for MSH2-dependent genomic stability. J. Biol. Chem. 271:7285-7288[Abstract/Free Full Text].

KARTHIKEYAN, R., E. J. VONARX, A. F. L. STRAFFON, M. SIMON, and G. FAYE et al., 2000  Evidence from mutational specificity studies that yeast DNA polymerases and replicate different DNA strands at an intracellular replication fork. J. Mol. Biol. 299:405-419[Medline].

KESTI, T., K. FLICK, S. KERANEN, J. E. SYVAOJA, and C. WITTENBERG, 1999  DNA polymerase catalytic domains are dispensable for DNA replication, DNA repair, and cell viability. Mol. Cell 3:679-685[Medline].

KROUTIL, L. C., K. REGISTER, K. BEBENEK, and T. A. KUNKEL, 1996  Exonucleolytic proofreading during replication of repetitive DNA. Biochemistry 35:1046-1053[Medline].

KUNKEL, T. A., 1992  Biological asymmetries and the fidelity of eukaryotic DNA replication. Bioessays 14:303-308[Medline].

KUNKEL, T. A. and K. BEBENEK, 2000  DNA replication fidelity. Annu. Rev. Biochem. 69:497-529[Medline].

KUNKEL, T. A. and A. SONI, 1988  Mutagenesis by transient misalignment. J. Biol. Chem. 263:14784-14789[Abstract/Free Full Text].

LEA, D. E. and C. A. COULSON, 1949  The distribution of the numbers of mutants in bacterial populations. J. Genet. 49:264-285.

LICHTEN, M., R. H. BORTS, and J. E. HABER, 1987  Meiotic gene conversion and crossing over between dispersed homologous sequences occur frequently in Saccharomyces cerevisiae. Genetics 115:233-246[Abstract/Free Full Text].

MARSISCHKY, G. T., N. FILOSI, M. F. KANE, and R. KOLODNER, 1996  Redundancy of Saccharomyces cerevisiae MSH3 and MSH6 in MSH2-dependent mismatch repair. Genes Dev. 10:407-420[Abstract/Free Full Text].

MODRICH, P. and R. LAHUE, 1996  Mismatch repair in replication fidelity, genetic recombination and cancer biology. Annu. Rev. Biochem. 65:101-133[Medline].

MORRISON, A. and A. SUGINO, 1994  The 5' to 3' exonucleases of both DNA polymerases and participate in correcting errors of DNA replication in Saccharomyces cerevisiae. Mol. Gen. Genet. 242:289-296[Medline].

MORRISON, A., J. B. BELL, T. A. KUNKEL, and A. SUGINO, 1991  Eukaryotic DNA polymerase amino acid sequence required for 3' to 5' exonuclease activity. Proc. Natl. Acad. Sci. USA 88:9473-9477[Abstract/Free Full Text].

MORRISON, A., A. L. JOHNSON, L. H. JOHNSTON, and A. SUGINO, 1993  Pathway correcting DNA replication errors in Saccharomyces cerevisiae. EMBO J. 12:1467-1473[Medline].

SCHAAPER, R. M., 1993  Base selection, proofreading, and mismatch repair during DNA replication in Escherichia coli. J. Biol. Chem. 268:23762-23765[Abstract/Free Full Text].

SHCHERBAKOVA, P. V. and Y. I. PAVLOV, 1996  3' to 5' exonucleases of DNA polymerases and correct base analog induced DNA replication errors on opposite DNA strands in Saccharomyces cerevisiae. Genetics 142:717-726[Abstract].

SHERMAN, F., 1991  Getting started with yeast. Methods Enzymol. 194:3-20[Medline].

STREISINGER, G., Y. OKADA, J. EMRICH, J. NEWTON, and A. TSUGITA et al., 1966  Frameshift mutations and the genetic code. Cold Spring Harbor Symp. Quant. Biol. 31:77-84[Medline].

SUGINO, A., 1995  Yeast DNA polymerases and their role at the replication fork. Trends Biochem. Sci. 20:319-323[Medline].

TOFT, N. J., D. J. WINTON, J. KELLY, L. A. HOWARD, and M. DEKKER et al., 1999  Msh2 status modulates both apoptosis and mutation frequency in the murine small intestine. Proc. Natl. Acad. Sci. USA 96:3911-3915[Abstract/Free Full Text].

TRAN, H. T., J. D. KEEN, M. KRICKER, M. A. RESNICK, and D. A. GORDENIN, 1997  Hypermutability of homonucleotide runs in mismatch repair and DNA polymerase proofreading yeast mutants. Mol. Cell. Biol. 17:2859-2865[Abstract].

TRAN, H. T., D. A. GORDENIN, and M. A. RESNICK, 1999  The 3'-5' exonucleases of DNA polymerases and and the 5'-3' exonuclease Exo1 have major roles in postreplication mutation avoidance in Saccharomyces cerevisiae. Mol. Cell. Biol. 19:2000-2007[Abstract/Free Full Text].

WANG, T.-F., N. KLECKNER, and N. HUNTER, 1999  Functional specificity of MutL homologs in yeast: evidence for three Mlh1-based heterocomplexes with distinct roles during meiosis in recombination and mismatch correction. Proc. Natl. Acad. Sci. USA 96:13914-13919[Abstract/Free Full Text].

This article has been cited by other articles:

				R. N. Venkatesan, P. M. Treuting, E. D. Fuller, R. E. Goldsby, T. H. Norwood, T. A. Gooley, W. C. Ladiges, B. D. Preston, and L. A. Loeb Mutation at the Polymerase Active Site of Mouse DNA Polymerase {delta} Increases Genomic Instability and Accelerates Tumorigenesis Mol. Cell. Biol., November 1, 2007; 27(21): 7669 - 7682. [Abstract] [Full Text] [PDF]

				A. L. Abdulovic and S. Jinks-Robertson The in Vivo Characterization of Translesion Synthesis Across UV-Induced Lesions in Saccharomyces cerevisiae: Insights Into Pol{zeta}- and Pol{eta}-Dependent Frameshift Mutagenesis Genetics, March 1, 2006; 172(3): 1487 - 1498. [Abstract] [Full Text] [PDF]

				B. Trouiller, D. G. Schaefer, F. Charlot, and F. Nogue MSH2 is essential for the preservation of genome integrity and prevents homeologous recombination in the moss Physcomitrella patens Nucleic Acids Res., January 5, 2006; 34(1): 232 - 242. [Abstract] [Full Text] [PDF]

				M.-E. Huang, A.-G. Rio, M.-D. Galibert, and F. Galibert Pol32, a Subunit of Saccharomyces cerevisiae DNA Polymerase {delta}, Suppresses Genomic Deletions and Is Involved in the Mutagenic Bypass Pathway Genetics, April 1, 2002; 160(4): 1409 - 1422. [Abstract] [Full Text] [PDF]