The Role of the Mismatch Repair Machinery in Regulating Mitotic and Meiotic Recombination Between Diverged Sequences in Yeast

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

Nonidentical recombination substrates recombine less efficiently than do identical substrates in yeast, and much of this inhibition can be attributed to action of the mismatch repair (MMR) machinery. In this study an intron-based inverted repeat assay system has been used to directly compare the rates of mitotic and meiotic recombination between pairs of 350-bp substrates varying from 82% to 100% in sequence identity. The recombination rate data indicate that sequence divergence impacts mitotic and meiotic recombination similarly, although subtle differences are evident. In addition to assessing recombination rates as a function of sequence divergence, the endpoints of mitotic and meiotic recombination events involving 94%-identical substrates were determined by DNA sequencing. The endpoint analysis indicates that the extent of meiotic heteroduplex DNA formed in a MMR-defective strain is 65% longer than that formed in a wild-type strain. These data are consistent with a model in which the MMR machinery interferes with the formation and/or extension of heteroduplex intermediates during recombination.

HOMOLOGOUS recombination involves the formation of heteroduplex DNA in which single strands of DNA derived from different parental duplexes are base-paired. The point at which the duplexes exchange pairing partners is referred to as a Holliday junction, and endonucleolytic cleavage of this junction can either maintain or reverse the linkage of markers that flank the region of heteroduplex DNA. Mismatches present in heteroduplex DNA are corrected by the postreplicative mismatch repair (MMR) machinery and such repair results in the genetic phenomenon of gene conversion. The concerted conversion of a contiguous series of potential mismatches constitutes a gene conversion tract, the length of which can be used as a minimal estimate of the extent of heteroduplex formed during recombination (AHN and LIVINGSTON 1986 ; JUDD and PETES 1988 ; SYMINGTON and PETES 1988 ; BORTS and HABER 1989 ; MEZARD et al. 1992 ; HARRIS et al. 1993 ; SWEETSER et al. 1994 ; PORTER et al. 1996 ; CHEN and JINKS-ROBERTSON 1998 ).

Although mismatches often are used to infer the nature of recombination intermediates, sequence divergence has been found uniformly to decrease recombination in bacterial species, yeast, and mammalian cells (for a review, see MODRICH and LAHUE 1996 ). Surprisingly, a single mismatch within a region of otherwise perfect identity is sufficient to inhibit transformation in Bacillus (CLAVERYS and LACKS 1986 ) or mitotic recombination in yeast (DATTA et al. 1997 ). In both bacteria and yeast, a log-linear relationship between the rate of recombination and the level of sequence divergence has been observed (ZAWADZKI et al. 1995 ; DATTA et al. 1997 ; VULIC et al. 1997 ). Such a relationship is consistent with the concept of a minimal efficient processing segment (MEPS; SHEN and HUANG 1986 ), which is defined as the length of perfect identity needed to efficiently initiate recombination. If one considers recombination substrates as a linear series of overlapping MEPS, then one would predict that the number of MEPS should be a linear function of substrate length and an exponential function of substrate identity.

Much of the limitation imposed on recombination between diverged (homeologous) sequences derives from action of the MMR system. Inactivation of a component(s) of the MMR system usually increases the rate of homeologous recombination, sometimes restoring it to a level comparable to the rate of recombination between identical sequences (RAYSSIGUIER et al. 1989 ; BORTS et al. 1990 ; ZAHRT et al. 1994 ; DE WIND et al. 1995 ; ZAWADZKI et al. 1995 ; DATTA et al. 1996 , DATTA et al. 1997 ; HUNTER et al. 1996 ; ZAHRT and MALOY 1997 ). The biochemistry of the MMR system has been best characterized in Escherichia coli, where the primary role of the system is to remove incorrect nucleotides incorporated during DNA synthesis (reviewed in MODRICH and LAHUE 1996 ). In E. coli, a MutS homodimer binds to the mismatched bases and MutH binds to a nearby hemi-methylated dam site, which is present in newly replicated DNA. A MutL homodimer acts as a "molecular matchmaker" to bring together the MutS and MutH proteins, which in turn activates a latent endonuclease activity of MutH. The unmethylated, newly synthesized DNA strand is then nicked, thus marking it for removal by helicases and exonucleases.

Mismatch recognition systems similar to the E. coli MutHLS system have been described in eukaryotes and have attracted much attention because defects have been associated with some forms of human hereditary cancer (reviewed in KOLODNER 1996 ). In yeast, multiple homologs of the E. coli MutS (Msh1-6p) and MutL (Pms1p, Mlh1-3p) proteins have been identified, with the major players in nuclear mismatch recognition being Msh2p, Msh3p, Msh6p, Pms1p, and Mlh1p (reviewed in CROUSE 1998 ). Msh2p dimerizes with either Msh3p or Msh6p, and each of these heterodimers recognizes specific types of mismatches. Pms1p likewise forms heterodimers with Mlh1p, and these heterodimers interact with the Msh2p-containing heterodimers in a manner analogous to the bacterial MutS-MutL interaction. Although the yeast Msh2 and Pms1/Mlh1 proteins appear to participate equally in the postreplicative repair of mismatched bases, their antirecombination activities are different. It has been shown in several studies that elimination of Msh2p stimulates recombination between diverged sequences to a greater extent than does elimination of Pms1p (CHAMBERS et al. 1996 ; DATTA et al. 1996 ). Methylation is not used as a strand discrimination signal in eukaryotes, so it is not surprising that no eukaryotic MutH homologs have been identified.

The antirecombination activity of the MMR machinery presumably derives from the recognition of mismatches present in heteroduplex recombination intermediates, but how the MMR machinery inhibits recombination once mismatches are detected is not clear. Based on studies in both bacteria (CLAVERYS and LACKS 1986 ; ZAHRT et al. 1994 ; ZAHRT and MALOY 1997 ) and yeast (ALANI et al. 1994 ), it has been proposed that mismatch recognition by the MMR machinery might trigger either helicase-driven unwinding of heteroduplex DNA or immediate resolution of the heteroduplex intermediate. According to such models, one might predict that the extent of heteroduplex formation should be greater in the absence of MMR than in its presence. WORTH et al. 1994 indeed have shown that both the rate and extent of in vitro RecA-catalyzed heteroduplex formation are reduced in the presence of purified MutS and MutL proteins. In yeast, ALANI et al. 1994 have argued that meiotic heteroduplex is longer in MMR-deficient cells than in wild-type cells, and we have found that mitotic gene conversion tracts in a msh2 msh3 strain are 50% longer than those in a wild-type strain (CHEN and JINKS-ROBERTSON 1998 ).

Homologous recombination in yeast occurs during both mitotic and meiotic cell divisions. In mitotically dividing cells recombination constitutes an important mechanism for repairing broken DNA molecules that arise as a result of random DNA damage. In meiosis, recombination also repairs broken DNA molecules, but the breaks are generated enzymatically at nonrandom sites (KEENEY et al. 1997 ). In contrast to the dispensable role recombination plays in yeast mitosis, meiotic recombination serves an essential function by providing a physical link between homologous chromosomes. This link ensures the attachment of homologs to opposite poles of the meiotic spindle and their subsequent disjunction (reviewed in ROEDER 1995 ). Yeast studies have demonstrated that the mitotic (SELVA et al. 1995 , SELVA et al. 1997 ; DATTA et al. 1996 , DATTA et al. 1997 ) and meiotic (BORTS et al. 1990 ; CHAMBERS et al. 1996 ; HUNTER et al. 1996 ) recombination barriers imposed by sequence divergence (see above) can be partially relieved if one or more of the MMR genes is inactivated. Unfortunately, it has not been possible to directly compare the published mitotic vs. meiotic data because of the very different assay systems that have been used. In addition, meiotic studies have not employed the systematic variation in substrate identity that has been used in some mitotic studies (DATTA et al. 1997 ).

The relative sequence identity requirements for mitotic vs. meiotic recombination is an interesting issue. On the one hand, one might expect the identity requirements of meiotic recombination to be more stringent than those of mitotic recombination in order to ensure that most interactions are allelic interactions between homologs rather than ectopic interactions between dispersed repeats. Ectopic interactions have the potential to generate genome rearrangements, which, if they occur in meiosis, can directly impact gamete viability as well as the fitness of progeny. On the other hand, because recombination is necessary for proper homolog disjunction in meiosis, one might expect meiotic recombination to forego the stringent homology requirements of mitotic recombination in order to guarantee that at least one crossover occurs between each pair of homologs. A fair comparison of mitotic and meiotic sequence identity requirements, as well as the role of the MMR machinery in enforcing these requirements, necessitates the use of the same system to measure the rates of both types of events. In this study an intron-based inverted repeat assay system was used to measure and directly compare the rates of mitotic vs. meiotic recombination between pairs of nonidentical substrates. In addition, mitotic and meiotic conversion tract endpoints in wild-type vs. MMR-defective strains were determined to ascertain the impact of MMR proteins on the formation of recombination intermediates.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media and growth conditions:
Yeast strains were grown nonselectively in YEP medium (1% yeast extract, 2% Bacto-peptone; 2.5% agar for plates) supplemented with either 2% glycerol and 2% ethanol (YEPGE) or 2% dextrose (YEPD). Synthetic complete medium (SHERMAN 1991 ) supplemented with 2% galactose, 2% glycerol, and 2% ethanol but deficient in histidine (SCGGE-his) was used to select His⁺ recombinants. To select Ura^- segregants of Ura⁺ strains, 5-fluoroorotic acid (5-FOA) was added to minimal synthetic medium supplemented with appropriate nutritional requirements (BOEKE et al. 1987 ). For meiotic experiments, diploid strains were pregrown in YEPD overnight followed by sporulation in 2% KOAc supplemented with 10 µg/ml histidine. Both vegetative growth and sporulation were conducted at 30°.

E. coli strains used for plasmid manipulations were grown at 37° in LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl; 1.5% agar for plates) supplemented with 150 µg/ml ampicillin as appropriate.

Plasmid constructions:
All inverted repeat (IR) constructs were contained on a pRS306-based plasmid (SIKORSKI and HIETER 1989 ), which has URA3 as a selectable marker. An outline of the plasmid constructions and the alignments of the IR substrates used in this study are presented in Figure 1 and Figure 2, respectively. Detailed descriptions of the plasmid constructions as well as the IR substrates (with the exception of the cß2a/cß2a-4mmA substrates; see below) can be found in DATTA et al. 1996 , DATTA et al. 1997 . pSR615 contains the cß2a/cß2a-4mmA IR substrates, which were derived from pAB96 (obtained from G. F. Crouse). pAB96 contains chicken ß tubulin isoform 2 (cß2a) cDNA sequences in the 5' recombination cassette, cß2a-4mmA sequences in the 3' cassette, and LEU2 as a selectable marker. cß2a-4mmA differs from cß2a by four evenly spaced base substitutions that were introduced by site-directed mutagenesis. pSR615 was constructed by ligating an AatII/NgoMI IR-containing fragment from pAB96 to AatII/NgoMI-digested pRS306, thereby changing the selectable marker from LEU2 to URA3.

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Construction of IR substrates. The pGAL-HIS3::intron construct contained on plasmid pSR266 is shown at the top. Open boxes correspond to HIS3 sequences, solid boxes to artificial intron sequences, and gray boxes to cß sequences; boxes are not to scale. The positions of the oligonucleotides used as PCR primers are indicated by arrows numbered as follows: 1, HIS3-702F; 2, HIS3-1751F; 3, HIS3-765R; 4, T3. The recombinant cß segment within the intron was amplified using primers 1 and 3; the recombinant segment downstream of the reconstituted HIS3::intron gene was amplified using primers 2 and 4.

View larger version (55K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Alignment of recombination substrates derived from cß cDNA sequences. (A) cß sequences were numbered as in the GenBank files (see DATTA et al. 1997 for details). For substrates containing one to four mismatches, the nature and position of the mismatches are indicated as a subscript. Each potential mismatch (mm) between a given pair of sequences is indicated by a vertical line. All sequences are ~350 bp. (B) Alignment of cß2/cß2-21 mm sequences. Sequences of perfect identities are boxed and shaded in gray. Although we refer to positions within the aligned sequences using the numbering system shown, it should be noted that the cß2a/cß2a-21mm homology does not actually begin until position 7.

Yeast strain constructions:
A complete list of strains used in this study is given in Table 1. All diploid strains are isogenic and were constructed by mating derivatives of haploid strain SJR216 with derivatives of either SJR231 or SJR328. A pms1 derivative of each parental haploid strain was constructed by a standard two-step gene transplacement method (ROTHSTEIN 1991 ) using BstXI-digested pSR211 (DATTA et al. 1996 ). A msh2::hisG derivative of each parental haploid was constructed by one-step gene disruption (ROTHSTEIN 1991 ) using AatII/XbaI-digested GC1914 (msh2::hisG-URA3-hisG plasmid obtained from G. F. Crouse). Ura⁺ transformants were selected following transformation with either pSR211 or GC1914. Transformants were purified nonselectively on YEPD and then plated on 5-FOA medium to identify Ura^- segregants. The MSH3 allele of SJR231 msh2::hisG was converted to msh3::hisG by one-step gene disruption using EcoRI-digested pEN33 (msh3::hisG-URA3-hisG plasmid; see DATTA et al. 1996 ) as described above. All gene disruptions were confirmed by PCR.

Plasmids containing identical or mismatched IR constructs were transformed into the isogenic haploids SJR328, GCY121 (SJR231 msh2::hisG msh3::hisG), GCY128 (SJR231 pms1), or SJR626 (SJR328 pms1). Plasmids were targeted to the URA3 locus by digestion with StuI and integration of a single copy of each plasmid was confirmed by Southern analysis.

Measuring mitotic and meiotic recombination rates:
Diploid strains were created immediately before rate measurement experiments in order to avoid the accumulation of recessive lethal mutations. Two independent diploids, each derived by mating two independently constructed haploid parents, were used for rate determinations. Diploids were constructed by mixing appropriate haploids on YEPD medium; after 5 hr, the mating mixtures were streaked onto medium selective for diploids. Two-day-old diploid colonies were used directly to inoculate 5 ml of YEPD medium and cultures were grown overnight to a density of ~2 x 10⁸ cells/ml. Cells were washed with 5 ml of sterile H₂O and resuspended in 1 ml of sterile H₂O. For mitotic rate determinations, aliquots of appropriate dilutions were plated in duplicate on YEPD medium to determine the total viable cell count and on SCGGE-his medium to select for His⁺ recombinants. Colonies on YEPD and SCGGE-his plates were counted 2 and 5 days, respectively, after plating. Data from 12 or more cultures (6 from each independent diploid) of each strain were used to calculate the mitotic recombination rate by the method of the median (LEA and COULSON 1949 ).

For meiotic rate determinations, portions of the vegetatively grown cultures used to measure mitotic recombination rates were sporulated at a density of 1–2 x 10⁷ cells/ml. Random spores were prepared by treating sporulated cultures with ß-mercaptoethanol, followed by glusulase treatment to digest the ascal wall and kill vegetative cells (DAVIDOW and BYERS 1984 ). The cultures were then sonicated to disperse the spores, and aliquots of appropriate dilutions were plated nonselectively on YEPD medium and selectively on SCGGE-his medium. Colonies arising on YEPD and SCGGE-his plates were counted 2 and 6 days, respectively, after plating. The meiotic His⁺ recombination rate for a particular strain was calculated by averaging the rates determined for 6–8 independent cultures (usually half from each independent diploid). All standard deviations were <35% of the rates in wild-type strains and <50% of the rates in the MMR-defective strains.

Generating independent recombinants for mapping conversion tract endpoints:
One-ml cultures, each inoculated using a different colony, were grown nonselectively in YEPGE medium to generate mitotic recombinants for molecular analysis. Cells were washed with sterile H₂O, resuspended in 200 µl of sterile H₂O, and 100 µl were plated on SCGGE-his medium to select for His⁺ recombinants. Only one colony was taken from each culture to ensure that all mitotic recombinants analyzed were of independent origin. Because recombinants generated in meiosis do not divide before selective plating, each meiotic recombinant derived from a given culture can be assumed to be of independent origin. Meiotic recombinants of each strain were, therefore, obtained directly from the SCGGE-his plates used to determine meiotic recombination rates.

Molecular analysis of recombinants:
Genomic DNA was extracted by glass bead lysis (HOFFMAN and WINSTON 1987 ) from each recombinant and used as a template for PCR amplification. Recombination products were amplified individually using primers annealing to sequences flanking the recombination substrates (Figure 1). The recombination product located within the intron was amplified using primers HIS3-702F (5'-GTTTCTGGACCATATG) and HIS3-765R (5'-GCACTCAACGATTAG). The recombination product located downstream of the reconstituted HIS3::intron gene was amplified using primers HIS3-1751F (5'-GATGGCAAACATGTC) and T3 (5'-TGATGTCGGCGATATAGG). PCR products were purified using Qiaquick Spin Columns (QIAGEN, Chatsworth, CA) and were used directly as templates for DNA sequencing. FAI (5'-ATGGACTAAAGGAGGCT) and T3 were used as sequencing primers for the intron and downstream recombination products, respectively. All sequencing reactions were carried out using ABI Prizm Dye Terminator Cycle Sequencing Ready Reaction Kits and were run on an ABI Prizm 377XL DNA sequencer (PE Applied Biosystems).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The intron-based inverted repeat recombination assay system:
The assay system used to examine the effects of sequence divergence on recombination was derived from a galactose-inducible HIS3 gene containing an artificial intron (HIS3::intron). As illustrated in Figure 1, replacement of the 3' or 5' half of HIS3::intron with a 350-bp recombination substrate created a 5' or 3' recombination cassette, respectively. All recombination substrates were derived from chicken ß-tubulin cDNA (cß) sequences and substrate pairs varied in identity from 82% to 100% (Figure 2). Juxtaposition of a 5' and 3' cassette in reverse orientation creates an IR construct with the recombination substrates flanking the 3' half of the HIS3::intron gene. Recombination between the substrates via either intrachromatid crossover or sister chromatid conversion flips the intervening HIS3::intron sequences (the "invertible segment"), thus reconstituting a full-length HIS3 gene with a complete intron containing one of the two recombination products. The other recombination product is located distal to the intact HIS3::intron gene. Because the recombinant cß sequences within the HIS3::intron gene are spliced out of the primary transcript and do not impact the gene product, there are no functional constraints on either the recombination substrates or the recombination products. It should be noted that neither intrachromatid gene conversion nor sister chromatid crossover produces His⁺ recombinants. Intrachromatid gene conversion does not reorient the 3' HIS3::intron segment and so recombinants are His^-; a sister chromatid crossover produces acentric and dicentric chromosomes and hence inviable His⁺ progeny.

We previously used the intron-based assay system to demonstrate that mitotic recombination between diverged sequences in yeast is regulated in large part by the MMR machinery (DATTA et al. 1996 , DATTA et al. 1997 ). These studies, however, provided no information concerning the antirecombination role of the MMR machinery in meiotic recombination. To specifically address this issue and to directly compare the identity requirements of mitotic vs. meiotic recombination, MAT strains containing IR constructs were mated with appropriate MATa haploid strains to create wild-type, MSH2-defective (msh2/msh2 MSH3/msh3), or PMS1-defective (pms1/pms1) diploid strains. The MSH2-defective and PMS1-defective diploid strains will hereafter be referred to as msh2 and pms1 strains, respectively. It should be noted that all diploids were heterozygous for the IR plasmid, thus precluding the production of His⁺ recombinants by interchromosomal interactions.

Recombination rates between identical and mismatch-containing substrates:
Mitotic and meiotic recombination rates were inferred in each strain by measuring the rates of His⁺ prototroph formation by fluctuation analysis and random spore analysis, respectively, and these rates are presented in Table 2. Recombination rates between 100% identical cß2a sequences were measured in the wild-type, msh2, and pms1 strains, and these rates were used as a normalization standard when assessing the effects of mismatches on recombination in the presence or absence of MMR proteins. Mitotically, the cß2a/cß2a 100% substrates recombined at a rate of ~1 x 10^-6 in the wild-type, msh2, and pms1 strains. This similarity in recombination rates was unexpected because we consistently have found a two- to threefold elevation in recombination rates between identical sequences in msh2 strains relative to MMR-competent strains (DATTA et al. 1997 ; A. BAYLISS, M. HENDRIX, G. F. CROUSE and S. JINKS-ROBERTSON, unpublished results). Although we do not understand the reason for this discrepancy, we speculate that it could reflect either a haploid/diploid effect (all previous studies were done with haploid strains) or strain background differences (one of the haploid parents used in this study is the same as that used in other studies, but the other is unrelated).

The rate of meiotic recombination between the cß2a/cß2a 100% substrates was ~100-fold greater than the corresponding mitotic rate in the wild-type, msh2, and pms1 strains. It should be noted, however, that the induction of meiotic recombination is usually assessed using recombination frequencies rather than rates. Whereas the meiotic recombination frequency is equal to meiotic rate (all events occur in a single generation), the mitotic frequency is generally higher than mitotic rate because of the random occurrence of mitotic recombination over several generations. With our substrates, the 100-fold difference between mitotic and meiotic recombination rates translates to approximately a 10-fold difference in mitotic vs. meiotic recombination frequencies. This is considerably less than the several hundred-fold meiotic induction generally observed with allelic sequences, and may be related to the inherent bias for intrachromosomal events in mitosis (FABRE et al. 1984 ; KADYK and HARTWELL 1992 ; JINKS-ROBERTSON et al. 1993 ) vs. interchromosomal events in meiosis (HABER et al. 1984 ; GAME et al. 1989 ; SCHWACHA and KLECKNER 1997 ). It can be estimated that ~10% of "meiotic" events involving the IR substrates are likely of mitotic origin, but we do not believe that this 10% significantly impacts the meiotic data.

We reported previously that a single mismatch within the 350-bp IR substrates was sufficient to reduce mitotic recombination 4-fold in a wild-type haploid strain, and this reduction was entirely dependent on the MMR machinery (DATTA et al. 1997 ). Mitotic recombination rates in diploid strains were measured for two pairs of substrates whose sequences differ by a single nucleotide: 1mmA and 1mmB cß2a sequences differing at positions 884 (A884G) and 883 (G883C), respectively. Although the 1mmA and 1mmB substrates differ in the types of mismatches that potentially can be formed in heteroduplex recombination intermediates (A/C or G/T mismatch for 1mmA; G/G or C/C mismatch for 1mmB), their mitotic recombination rates in a wild-type background were decreased to similar extents relative to the 100% control rate (6.0-fold for 1mmA and 4.4-fold for 1mmB). In contrast to the approximately 5-fold effect observed mitotically, the mismatches in the 1mmA and 1mmB substrates reduced meiotic recombination only 2-fold and 1.8-fold, respectively, relative to the 100% control substrates. This indicates that a single potential mismatch has less impact on meiotic than on mitotic recombination rates. Elimination of MSH2 restored the mitotic and meiotic recombination rates of 1mmA to those of the 100% control substrates. Although deletion of the PMS1 gene also increased the mitotic recombination rate between the 1mmA substrates, the pms1 rate for the mismatched substrates was lower than the 100% control rate. Meiotic recombination between the 1mmA substrates was not impacted by elimination of Pms1p.

To examine the effects of multiple mismatches on mitotic and meiotic recombination, substrates containing three or more mismatches were introduced into the wild-type and MMR-defective diploid strains. As observed for the 1mm substrates, three mismatches reduced mitotic recombination rates more than meiotic rates (6.7-fold vs. 3.5-fold) in a wild-type background, and both reductions were completely dependent on the MMR machinery. When the recombination substrates differed by four or more nucleotides, however, meiotic recombination rates in a wild-type background were reduced just as much as the mitotic rates (Table 2). Although elimination of Msh2p or Pms1p increased both mitotic and meiotic recombination rates, the rates for the 94%, 91%, 85%, and 82% identical substrates were not equivalent to those observed with the 100% control substrates. The MMR-independent decrease in recombination rates with these lower levels of sequence identity presumably reflects an inability of the recombination machinery to efficiently initiate recombination between these sequences (DATTA et al. 1997 ). In contrast to the similar inhibitory effects of sequence divergence on mitotic and meiotic recombination rates between the 94%, 91%, 85%, and 82% identical substrates in MMR-competent strains, it should be noted that, in MMR-defective cells, meiotic recombination between these substrates was consistently impacted more by a decrease in sequence identity than was mitotic recombination. Mitotic recombination rates between the 85% or 82% identical substrates, for example, were reduced an average of 16-fold in the msh2 strains and 63-fold in the pms1 strains. The corresponding meiotic recombination rates between the 85% or 82% identical substrates were reduced an average of 50-fold in the msh2 strains and 180-fold in the pms1 strains relative to the 100% control substrates.

Mitotic and meiotic gene conversion tract endpoints in wild-type and msh2 strains:
One approach to addressing the mechanism of the MMR-associated inhibition of recombination is to determine whether the formation of recombination intermediates is altered by the MMR machinery. Because gene conversion tracts are generally assumed to be a direct reflection of the heteroduplex DNA intermediate formed during recombination, 94% identical substrates were used to map the endpoints of conversion tracts in mitotic and meiotic His⁺ recombinants derived from both wild-type and msh2 diploid strains. Conversion tract endpoints were determined by individually sequencing the recombinant cß segments flanking the invertible HIS3::intron segment.

As illustrated in Figure 2, the cß2a and cß2a-21mm sequences differ at 21 positions, thereby dividing the substrates into 21 intervals of perfect identity (two mismatches are adjacent to each other). A given mismatch was considered to have been converted if the corresponding nucleotides in the recombinant cß2a sequences were identical. A gene conversion tract encompasses a series of contiguous mismatches and has endpoints in two discrete intervals. An endpoint was assigned to a given interval (intervals are identified by the positions of the flanking mismatches) if the mismatch defining one side of the interval was converted but the mismatch defining the other side was not. If one assumes that gene conversion tracts start and end at random, then the number of endpoints contained in a given identity interval should be directly proportional to the length of that interval. Interval 318–350, for example, contains 31 bp of perfect identity, which constitutes ~10% of the 322 bp of perfect identity shared between the cß2a/cß2a-21mm substrates. One would predict, therefore, that this interval should contain ~10% of all conversion tract endpoints. The experimentally determined conversion tract endpoint distributions were compared to the expected distribution by subtracting the percentage of expected endpoints in each identity interval from the percentage of observed endpoints. This yields positive and negative percentages which, when plotted, indicate an excess or deficit of endpoints, respectively.

Gene conversion tracts were determined for 30 mitotic His⁺ recombinants generated in the wild-type strain. Twenty-eight of the recombinants had continuous conversion tracts with all mismatches converted in the same direction. The other two recombinants had no detectable conversion events and, therefore, were assumed to have both endpoints in the same interval. The expected and observed distributions of conversion tract endpoints are presented in Table 3 and are compared graphically in Figure 3A. Several of the 3' intervals (e.g., intervals 318–350 and 282–309) have a notable excess of endpoints, indicating that the ends of the substrates proximal to the invertable HIS3::intron segment may be preferred sites for initiating and/or resolving recombination intermediates.

View larger version (15K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Conversion tract endpoint distributions. The percentages of endpoints expected in each interval (see Table 3 for the interval numbers) were subtracted from the observed percentages, and the residual values were plotted (solid circles). Positive or negative deviations of observed from expected percentages indicate an excess or deficit of endpoints, respectively. Vertical bars correspond to standard deviations (SDs), each of which was approximated by taking the square root of the number of endpoints observed in a given interval. The SDs thus obtained were converted to percentages, and these percentages were added to and subtracted from the corresponding percentage deviations. The SD for interval 21 in B is off the y-axis scale.

To directly compare mitotic and meiotic conversion tracts, conversion tracts were determined for 33 meiotic His⁺ recombinants isolated from the same wild-type strain. Of these conversion tracts, 27 were continuous and asymmetric and 3 had no mismatches converted. The remaining 3 recombinants contained either interrupted conversion tracts or continuous but bidirectional (symmetric) tracts. Because of their complexity, these latter tracts were not included in the determination of endpoint distributions or in calculations of conversion tract lengths (see below). The distribution of the meiotic conversion tract endpoints is presented in Table 3 and graphically compared to the expected distribution in Figure 3B. Strikingly, the most 3' interval, interval 318–350, contained 33% of all endpoints whereas only 10% were predicted to be in this interval on the basis of its length. Most of the remaining identity intervals exhibited a mild deficit of endpoints. A comparison of the distributions in Figure 3A and Figure B, indicates that the moderate mitotic clustering of endpoints at the 3' end of the substrates is exaggerated in the corresponding meiotic recombinants.

In the absence of a functional MMR system, the mismatches present in heteroduplex molecules formed during recombination should be segregated at the next round of DNA replication, thus producing the equivalent of a gene conversion tract. Sixty-three mitotic His⁺ recombinants derived from the msh2 strain were sequenced to estimate the extent of heteroduplex formation. Fifty recombinants had continuous asymmetric conversion tracts, 6 had no mismatches converted, and 7 had complex conversion events. This class distribution is not statistically different from that of the mitotic conversion tracts in wild-type cells (P > 0.1 by ² contingency test). The distribution of conversion tract endpoints is presented in Table 3 and graphically compared to the expected distribution in Figure 3C. In contrast to the weak 3' clustering of mitotic conversion tract endpoints evident in wild-type cells (Figure 3A), endpoints appeared to be more or less randomly distributed in the MMR-defective cells.

Thirty-three meiotic His⁺ recombinants derived from the msh2 strain also were analyzed. Twenty-six recombinants had continuous asymmetric conversion tracts, 4 had no mismatches converted, and 3 had complex tracts. This class distribution is the same as that of the meiotic recombinants isolated from a wild-type strain (P > 0.9 by ² contingency test). The meiotic conversion tract endpoint distribution is presented in Table 3 and is compared graphically to the expected distribution in Figure 3D. In contrast to the very striking clustering of meiotic endpoints in interval 318–350 in wild-type cells (Figure 3B), there was no evident clustering of meiotic endpoints in MMR-defective cells.

Mitotic and meiotic conversion tract lengths in wild-type and msh2 strains:
In the assay system used here, reorientation of the HIS3::intron segment between the recombination substrates is required in order for a recombinant to be His⁺ (Figure 1). Reorientation involves interactions between the flanking cß segments and can occur either by intrachromatid crossover or by sister chromatid conversion. As illustrated in Figure 4, an intrachromatid crossover requires pairing of two substrates on the same DNA molecule, whereas sister chromatid conversion involves pairing of two pairs of substrates, one on either side of the invertible segment.

View larger version (13K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Models of IR recombination. Inversion of a segment of DNA between inverted repeats can occur by either (A) intrachromatid crossover or (B) sister chromatid gene conversion. The open and solid boxes correspond to the IR substrates that flank the invertible HIS3::intron segment, which is represented by a loop with an arrow indicating orientation. Open circles denote centromeres. It should be noted that if the internally paired chromatid in A is straightened, it is identical to the top chromatid in B in terms of the recombination endpoints. Thus, although the conversion tract in A would be calculated to be quite short, a conversion tract with identical endpoints would be relatively long if one assumes the model in B.

Given the same conversion tract endpoints, the calculation of conversion tract length is very different depending on whether a recombination event occurs through the intrachromatid crossover pathway vs. the sister chromatid conversion pathway (see Figure 4). Although intrachromatid crossover and sister chromatid conversion are genetically indistinguishable, we have argued previously that most of the His⁺ recombinants selected by our inverted repeat system arise via the sister chromatid conversion pathway (CHEN and JINKS-ROBERTSON 1998 ). Using the experimentally determined conversion tract endpoints for each mitotic or meiotic recombinant, the corresponding conversion tract length was calculated according to the sister chromatid conversion model. As for the conversion tract endpoint analysis, only recombinants with continuous conversion tracts or no conversion tracts were included in the calculation of conversion tract lengths. It should be noted that a "no conversion" event corresponds to a sister chromatid conversion in which the endpoints on either side of the HIS3::intron invertible segment are in the same interval (e.g., an event that initiates in interval 191–219 on one side of the invertible segment, proceeds through the invertible segment, and ends in interval 191–219 on the other side of the invertible segment).

Table 4 presents the mean values of the minimal, maximal, and average conversion tract lengths, as well as the average number of mismatches converted. The mitotic gene conversion tracts in the wild-type diploid strain averaged 275 bp, which agrees very well with the 280-bp average length reported in haploid cells (CHEN and JINKS-ROBERTSON 1998 ). In msh2 diploid cells, the average mitotic tract length was 312 bp, which is slightly, but not significantly, longer than the tract length in wild-type diploid cells. This is in contrast to the statistically significant lengthening of tracts (from 275 bp to 385 bp) observed previously by us in a msh2 msh3 haploid strain (CHEN and JINKS-ROBERTSON 1998 ). One possible reason for the tract length difference in the msh2 diploid cells vs. the msh2 msh3 haploid cells is that Msh3p might be involved specifically in regulating conversion tract length independently of Msh2p. We tested this possibility by analyzing conversion tracts in a msh2 haploid strain isogenic to the previously used msh2 msh3 haploid (data not shown). The average conversion tract length in the msh2 haploid was 349 bp, which is significantly different from the 275 bp length in wild-type haploid cells (P < 0.05 by Student's t-test) but not significantly different from the 385-bp average length in the msh2 msh3 haploid cells (P = 0.33 by Student's t-test). We thus are left with the speculation that the discrepancy between our haploid and diploid results may reflect a haploid/diploid difference in recombination or could reflect the fact that the haploid and diploid strains are not isogenic.

The average length of meiotic conversion tracts was 204 bp in the wild-type diploid strain vs. 339 bp in the msh2 diploid strain. This 65% difference in meiotic tract lengths is statistically significant and suggests that the MMR machinery regulates the extent of heteroduplex formation during meiosis. Although mitotic tracts were longer than meiotic tracts in wild-type cells (280 bp vs. 230 bp) and meiotic tracts were longer than mitotic tracts in msh2 cells (361 bp vs. 312 bp), neither of these differences is statistically significant.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

An intron-based IR assay system was used to examine mitotic and meiotic recombination rates between identical and mismatched sequences in both MMR-competent and MMR-defective yeast strains. Recombination rates were measured in isogenic wild-type, msh2, and pms1 diploid strains, thus allowing a direct comparison of the impact of sequence divergence on mitotic vs. meiotic recombination events. In addition, recombination products derived from 94%-identical substrates were sequenced to estimate the extent of meiotic vs. mitotic heteroduplex formation in wild-type and msh2 strains. It should be noted that the substrates in all strains were present on only one copy of chromosome V, which limits detectable recombinants to intrachromosomal events and precludes the production of recombinants via recombination between homologs.

Recombination rates between mismatch-containing substrates in wild-type cells:
The rates of mitotic and meiotic recombination were measured between cß substrates varying in identity from 82% to 100% (see Figure 2 for substrate alignments). Recombination rates are given in Table 2, and all recombination rates obtained with a strain of a given genotype (wild-type, msh2, or pms1) were normalized to the rate obtained with 100%-identical cß2a control substrates. These normalized data are presented graphically in Figure 5 to more easily compare and contrast mitotic and meiotic recombination rates in the three strain backgrounds used. In a wild-type strain, the presence of one or three mismatches in the recombination substrates reduced mitotic recombination more than meiotic recombination (see inset in Figure 5). Although the mitotic vs. meiotic differences are subtle, they suggest that mismatches in meiotic heteroduplex intermediates are either recognized less efficiently than those in mitotic intermediates, or that once recognized, meiotic mismatches have a less negative impact on the overall recombination process. Alternatively, one could hypothesize that meiotic heteroduplex is shorter than mitotic heteroduplex and, therefore, less likely to include the mismatch(es) that trigger the MMR-associated antirecombination activity. The conversion tract length analysis indicates, however, that similar extents of heteroduplex are formed in mitosis and meiosis. With four or more mismatches, the normalized levels of mitotic and meiotic recombination rates for a given level of substrate identity were indistinguishable in a wild-type background. If one assumes that mismatches have a cumulative negative effect on recombination, then the probability of escaping the antirecombination activity of the MMR machinery will become essentially zero at some level of divergence, and the inclusion of additional mismatches will not further impact recombination. We suggest that the probability that a given mismatch triggers the antirecombination activity of MMR proteins is less in meiosis than in mitosis, but that the cumulative effect of multiple mismatches in both cases is to eventually trigger antirecombination with the same efficiency. The net result would be less inhibition of meiotic recombination than mitotic recombination by one or a few mismatches, but similar levels of inhibition by higher levels of sequence divergence.

View larger version (27K): [in this window] [in a new window] [Download PPT slide]   Figure 5. Relationship between recombination rates and sequence divergence. The natural logarithms of the normalized recombination rates in Table 2 are plotted as a function of the percentage of sequence divergence. The mitotic or meiotic normalized rates for each strain genotype were obtained by dividing each rate by the rate obtained with the 100% identical control substrates.

The differential effect of a very low level of sequence divergence on the overall efficiency of mitotic vs. meiotic recombination in wild-type cells makes biological sense if one considers the relative roles of recombination in mitosis vs. meiosis. Recombination is responsible for the repair of double-strand breaks (DSBs) generated by random DNA damage in mitosis and by nonrandom enzymatic cleavage in meiosis. Although mitotic ectopic recombination between nonhomologous chromosomes occurs at about the same level as that between allelic sequences on homologous chromosomes, intrachromosomal interactions (both sister chromatid and intrachromatid) are much more efficient than interchromosomal interactions in mitosis (FABRE et al. 1984 ; LICHTEN and HABER 1989 ; KADYK and HARTWELL 1992 ; JINKS-ROBERTSON et al. 1993 ). It has been argued that recombination between sister chromatids should be favored over interchromosomal interactions to ensure that mitotic DNA repair occurs with very high fidelity (sister chromatids are identical) and also to lower the occurrence of deleterious chromosome rearrangements. We suggest that the very stringent identity requirements imposed by the yeast MMR machinery in mitosis would discourage nonsister interactions and thereby further strengthen the natural bias for sister chromatid interactions. In meiosis, the bias for recombination between homologs vs. intrachromosomal interactions is reversed (HABER et al. 1984 ; GAME et al. 1989 ; SCHWACHA and KLECKNER 1997 ). This reversal of the bias may be related to formation of the synaptonemal complex between homologs, which could serve either to encourage allelic interactions, or to discourage both intrachromosomal interactions and interactions between nonhomologous chromosomes. Because meiotic recombination generally provides an essential physical link between homologous chromosomes that ensures their proper disjunction, maintaining a high efficiency of meiotic recombination in the presence of subtle sequence polymorphisms between homologs is crucial. We suggest that the relaxed identity requirements observed here with very small amounts of sequence divergence may aid in promoting the requisite nonsister interactions in meiosis. Alternatively, the differential effect of low levels of sequence divergence could reflect a switch from mismatch-triggered heteroduplex rejection in mitosis to mismatch-triggered heteroduplex repair in meiosis.

Recombination rates between mismatch-containing substrates in MMR-defective cells:
The effect of sequence divergence on mitotic vs. meiotic recombination was examined in two different types of MMR-defective diploid strains: msh2 and pms1. Whereas strains deleted for MSH2 (the MutS homolog essential for all mismatch repair) should have no mismatch binding activity, in vitro studies indicate that mismatch recognition can occur in the absence of the MutL homologs Pms1p and Mlh1p (HABRAKEN et al. 1997 ). To delineate the antirecombination role of the MMR machinery (factors other than the MMR machinery may affect recombination in a mismatch-dependent manner), an "MMR index" was calculated for each pair of substrates. The MMR index is defined here as the ratio of normalized recombination rate measured in the msh2 or pms1 strain to that measured in the wild-type, MMR-competent strain. As shown in Figure 6, both the msh2 and the pms1 mitotic and meiotic MMR indices increased initially with increasing sequence divergence, demonstrating that the MMR machinery actively inhibits recombination in a mismatch-dependent manner. Although mismatches had a cumulative negative effect on recombination in the presence of the MMR machinery, the increases in the MMR indices eventually plateau. As noted above, this behavior suggests that after a critical number of mismatches has been sensed by the MMR machinery, the probability of escaping the antirecombination activity of the MMR proteins is zero and the presence of additional mismatches is inconsequential. Although the plateau is not evident with substrates having <94% identity (21 mismatches), we have no substrates in the 4-to-21 mismatch range so the exact point(s) where the plateau occurs is not clear. It should be noted that these observations are very similar to those reported previously by DATTA et al. 1997 .

View larger version (16K): [in this window] [in a new window] [Download PPT slide]   Figure 6. MMR indices. The MMR index for a given level of substrate identity in a MMR-defective strain (either msh2 or pms1) was calculated by dividing the normalized recombination rate for the mutant strain by that obtained for the wild-type, MMR-competent strain. The normalized recombination rates are the rates obtained in a given genetic background divided by the rate for the 100% control substrates in the same genetic background.

The mitotic MMR index is greater than the meiotic index for every pair of substrates examined in both the msh2 and the pms1 strains. This observation suggests that the maximal antirecombination activity of the yeast MMR machinery is more efficient in mitosis than in meiosis. It may be, for example, that, regardless of the number of mismatches present, 5% of all meiotic heteroduplexes "escape" the antirecombination activity of the MMR machinery, whereas only 1–2% of all mitotic heteroduplexes escape this activity. Alternatively, a larger percentage of meiotic recombinants may be produced through a mechanism that does not involve extensive heteroduplex formation. Gap repair, for example, does not involve extensive heteroduplex formation. Another example of such a mechanism has been proposed by Resnick and colleagues (PORTER et al. 1996 ) and involves the production of recombination intermediates by replicative extension of an invading 3' end (thus generating true homoduplex) rather than by continued assimilation of the (mismatched) invading strand.

We previously reported that msh2 strains exhibit higher levels of mitotic recombination between diverged substrates than do isogenic pms1 strains (DATTA et al. 1996 ). The data summarized in Figure 5 and Figure 6 confirm the mitotic observation and demonstrate that meiotic recombination behaves similarly. The normalized values of mitotic and meiotic recombination between diverged substrates were uniformly higher in a msh2 background than in a pms1 background (Figure 5) and the MMR indices were consistently larger for the msh2 strains than for the pms1 strains (Figure 6). Thus, although the yeast MutL homolog Pms1p is apparently indispensable for the repair of DNA replication errors (reviewed in CROUSE 1998 ), it is not absolutely required for the antirecombination activity of the MMR machinery in either mitosis or meiosis. This is not to imply that the yeast MutL homologs have no antirecombination role, but rather that some inhibition of recombination occurs in their absence. We suggest that mismatch binding by MutS homologs in the absence of MutL homologs may be sufficient to impede recombination, whereas such binding is apparently unable to trigger repair of DNA replication errors. The specific requirement of the MutL homologs for the repair of replication errors may be related to their interactions with proliferating cell nuclear antigen (PCNA; JOHNSON et al. 1996 ; UMAR et al. 1996 ), which have been speculated to link MMR directly to the process of DNA replication.

For both the msh2 and pms1 strains, meiotic recombination was more negatively impacted by sequence divergence than was mitotic recombination (Figure 5). The slightly steeper slopes for the plots of the meiotic vs. mitotic data suggest that the MEPS for meiotic recombination may be longer than that for mitotic recombination. That is, a greater length of perfect identity may be required to successfully initiate meiotic recombination than to initiate mitotic recombination. Once initiation occurs, however, mismatches that become incorporated into a heteroduplex intermediate would become potential targets of the MMR-associated antirecombination activity.

Gene conversion tracts:
Recombination products derived using the 94%-identical substrates in both wild-type and msh2 strains were sequenced to determine the endpoints and estimate the lengths of mitotic and meiotic conversion tracts (Figure 3 and Table 4, respectively). It is assumed in analyses of this sort that conversion tracts are accurate representations of the extent of heteroduplex formed in MMR-competent cells. We acknowledge the possibility, however, that a conversion tract border may not always correspond to the extent of the underlying heteroduplex intermediate, but rather may reflect the border of mismatch correction. The most notable feature of conversion tracts in the wild-type strain is that intervals close to the invertible HIS3::intron segment showed an excess of endpoints, while the more distal intervals exhibited a deficit of endpoints. A similar, but more pronounced, pattern of endpoint distribution was observed previously using one of the wild-type haploid parents of the diploid used in this study (CHEN and JINKS-ROBERTSON 1998 ). In the earlier study, we systematically addressed possible explanations for the endpoint distribution by altering the substrates in defined ways. Most notably, reversal of the orientations of both substrates with respect to the invertible HIS3::intron segment reversed the distribution of endpoints so that intervals proximal (and formally distal) to the invertible HIS3::intron segment still contained an excess of endpoints. We concluded that the proximity of an interval to the HIS3::intron segment was the primary determinant of the proportion of endpoints likely to be located in that interval, regardless of the length or sequence of the interval. Because the same intron-based system was used in this study, we suggest that the distributions of mitotic and meiotic conversion tract endpoints in a wild-type strain (Figure 3A and Figure B, respectively) similarly reflect the relative proximity of intervals to the invertible HIS3::intron segment. In contrast to the results obtained with an MMR-competent strain, the distributions of mitotic and meiotic conversion tracts in a msh2 strain lacked an obvious clustering of endpoints (compare Figure 3A TO 3C, and Figure 3B TO 3D). A similar observation was made by us using both a haploid msh2 msh3 strain and a haploid pms1 strain (CHEN and JINKS-ROBERTSON 1998 ).

We argued previously (CHEN and JINKS-ROBERTSON 1998 ) that a sister chromatid conversion model could more readily account for the observed conversion tract endpoint distributions than could an intrachromatid crossover model (Figure 4). According to the sister chromatid conversion model, each recombination event that successfully reorients the invertible HIS3::intron segment must initiate within substrates on one side of the invertible segment, extend through the invertible segment, and resolve within the substrates on the other side of the invertible segment. The endpoint clustering evident in a wild-type strain thus reflects a bidirectional gene conversion gradient that extends in both directions from the selected site of conversion (the invertible segment). Using the assumption of sister chromatid conversion, the lengths of mitotic and meiotic gene conversion tracts were calculated for events occurring in both wild-type and msh2 strains (Table 4). As predicted by the endpoint distributions, these analyses indicated that mitotic and meiotic tracts were longer in a msh2 strain than in a wild-type strain, although the length difference was only significant for meiotic tracts.

The conversion tract data strongly implicate the MMR machinery in determining the distribution of conversion tract endpoints in recombination events involving diverged sequences. We suggest that these data are relevant to the antirecombination activity of MMR proteins and indicate that recombination intermediates are targeted by the MMR machinery. The recombination products observed in a wild-type strain presumably are those that escaped the antirecombination activity of the MMR machinery, but nevertheless may provide useful insight into the mechanism of antirecombination. We suggest that the gradient of conversion tract endpoints observed in wild-type cells reflects increasing obstruction of recombination by the MMR machinery as the heteroduplex formation progresses and more and more mismatches are incorporated. One possibility is that mismatches are recognized by the MMR machinery during or immediately after the formation of heteroduplex DNA. Such recognition might block further extension of the heteroduplex intermediate and thereby trigger helicase-driven unwinding of the intermediate.

Conclusions:
In MMR-competent cells, mitotic recombination was impacted more than meiotic recombination by a low level of sequence divergence, whereas both types of recombination were similarly impacted by higher levels of divergence. The differential effects of low levels of divergence on recombination were MMR-dependent and might possibly serve to strengthen the biases for mitotic intrachromosomal interactions vs. meiotic interchromosomal interactions. In MMR-defective cells, meiotic recombination was impacted more by sequence divergence than was mitotic recombination, suggesting that the MEPS may be slightly longer for meiosis than for mitosis. Recombination rates between nonidentical sequences were consistently higher in msh2 strains than in pms1 strains, indicating that the yeast MutS homologs can exert antirecombination activity in the absence of the MutL homolog Pms1p. For both mitotic and meiotic recombination events, low levels of sequence divergence impeded recombination predominantly via action of the MMR machinery whereas higher levels of sequence divergence impeded recombination via action of the MMR machinery plus an additional MMR-independent process. This latter process may reflect a requirement for a minimal length of perfect homology (the MEPS) to successfully initiate a recombination event.

	ACKNOWLEDGMENTS

This work was supported by National Institutes of Health grant GM-38464 (S.J.-R.). W.C. was supported in part by the Graduate Division of Biological and Biomedical Sciences and by a National Institutes of Health Medical Scientist Training grant (Emory University, Atlanta, GA).

Manuscript received September 9, 1998; Accepted for publication December 18, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AHN, B.-Y. and D. M. LIVINGSTON, 1986  Mitotic gene conversion lengths, coconversion patterns, and the incidence of reciprocal recombination in a Saccharomyces cerevisiae plasmid system. Mol. Cell. Biol. 6:3685-3693[Abstract/Free Full Text].

ALANI, E., R. A. G. REENAN, and R. D. KOLODNER, 1994  Interaction between mismatch repair and genetic recombination in Saccharomyces cerevisiae.. Genetics 137:19-39[Abstract].

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].

BORTS, R. H. and J. E. HABER, 1989  Length and distribution of meiotic gene conversion tracts and crossovers in Saccharomyces cerevisiae.. Genetics 123:69-80[Abstract/Free Full Text].

BORTS, R. H., W.-Y. LEUNG, W. KRAMER, B. KRAMER, and M. WILLIAMSON et al., 1990  Mismatch repair-induced meiotic recombination requires PMS1 gene product. Genetics 124:573-584[Abstract].

CHAMBERS, S., N. HUNTER, E. J. LOUIS, and R. H. BORTS, 1996  The mismatch repair system reduces meiotic homeologous recombination and stimulates recombination-dependent chromosome loss. Mol. Cell. Biol. 16:6110-6120[Abstract].

CHEN, W. and S. JINKS-ROBERTSON, 1998  Mismatch repair proteins regulate heteroduplex formation during mitotic recombination in yeast. Mol. Cell. Biol. 18:6525-6537[Abstract/Free Full Text].

CLAVERYS, J.-P. and S. A. LACKS, 1986  Heteroduplex deoxyribonucleic acid base mismatch repair in bacteria. Microbiol. Rev. 50:133-165[Free Full Text].

CROUSE, G. F., 1998 Mismatch repair systems in Saccharomyces cerevisiae, pp. 411–448 in DNA Damage and Repair, Volume 1: DNA Repair in Prokaryotes and Lower Eukaryotes, edited by J. A. NICKOLOFF and M. F. HOEKSTRA. Humana Press, Totowa, NJ.

DATTA, A., A. ADJIRI, L. NEW, G. F. CROUSE, and S. JINKS-ROBERTSON, 1996  Mitotic crossovers between diverged sequences are regulated by mismatch repair proteins in Saccharomyces cerevisiae.. Mol. Cell. Biol. 16:1085-1093[Abstract].

DATTA, A., M. HENDRIX, M. LIPSITCH, and S. JINKS-ROBERTSON, 1997  Dual roles for DNA sequence identity and the mismatch repair system in the regulation of mitotic crossing-over in yeast. Proc. Natl. Acad. Sci. USA 94:9757-9762[Abstract/Free Full Text].

DAVIDOW, L. S. and B. BYERS, 1984  Enhanced gene conversion and postmeiotic segregation in pachytene-arrested Saccharomyces cerevisiae. Genetics 106:165-183[Abstract/Free Full Text].

DE WIND, N., M. DEKKER, A. BERNS, M. RADMAN, and H. TE RIELE, 1995  Inactivation of the mouse Msh2 gene results in mismatch repair deficiency, methylation tolerance, hyperrecombination, and predisposition to cancer. Cell 82:321-330[Medline].

FABRE, F., A. BOULET, and H. ROMAN, 1984  Gene conversion at different points in the mitotic cycle of Saccharomyces cerevisiae.. Mol. Gen. Genet. 195:139-143[Medline].

GAME, J. C., K. C. SITNEY, V. E. COOK, and R. K. MORTIMER, 1989  Use of a ring chromosome and pulsed-field gels to study interhomolog recombination, double-strand DNA breaks and sister-chromatid exchange in yeast. Genetics 123:695-713[Abstract/Free Full Text].

HABER, J. E., P. C. THORBURN, and D. ROGERS, 1984  Meiotic and mitotic behavior of dicentric chromosomes in Saccharomyces cerevisiae.. Genetics 106:185-205[Abstract/Free Full Text].

HABRAKEN, Y., P. SUNG, L. PRAKASH, and S. PRAKASH, 1997  Enhancement of MSH2-MSH3-mediated mismatch recognition by the yeast MLH1-PMS1 complex. Curr. Biol. 7:790-793[Medline].

HARRIS, S., K. S. RUDNICKI, and J. E. HABER, 1993  Gene conversions and crossing over during homologous and homeologous ectopic recombination in Saccharomyces cerevisiae.. Genetics 135:5-16[Abstract].

HOFFMAN, C. S. and F. WINSTON, 1987  A ten-minute DNA preparation from yeast efficiently releases autonomous plasmids for transformation of Escherichia coli.. Gene 57:267-272[Medline].

HUNTER, N., S. R. CHAMBERS, E. J. LOUIS, and R. H. BORTS, 1996  The mismatch repair system contributes to meiotic sterility in an interspecific yeast hybrid. EMBO J. 15:1726-1733[Medline].

JINKS-ROBERTSON, S., M. MICHELITCH, and S. RAMCHARAN, 1993  Substrate length requirements for efficient mitotic recombination in Saccharomyces cerevisiae.. Mol. Cell. Biol. 13:3937-3950[Abstract/Free Full Text].

JOHNSON, R. E., G. K. KIVVALI, S. N. GUZDER, N. S. AMIN, and C. HOLM et al., 1996  Evidence for involvement of yeast proliferating cell nuclear antigen in DNA mismatch repair. J. Biol. Chem. 271:27987-27990[Abstract/Free Full Text].

JUDD, S. R. and T. D. PETES, 1988  Physical lengths of meiotic and mitotic gene conversion tracts in Saccharomyces cerevisiae.. Genetics 118:401-410[Abstract/Free Full Text].

KADYK, L. C. and L. H. HARTWELL, 1992  Sister chromatids are preferred over homologs as substrates for recombinational repair in Saccharomyces cerevisiae.. Genetics 132:387-402[Abstract].

KEENEY, S., C. N. GIROUX, and N. KLECKNER, 1997  Meiosis-specific DNA double-strand breaks are catalyzed by Spo11, a member of a widely conserved protein family. Cell 88:375-384[Medline].

KOLODNER, R., 1996  Biochemistry and genetics of eukaryotic mismatch repair. Genes Dev. 10:1433-1442