The msh2 Gene of Schizosaccharomyces pombe Is Involved in Mismatch Repair, Mating-Type Switching, and Meiotic Chromosome Organization

Abstract

We have identified in the fission yeast Schizosaccharomyces pombe a MutS homolog that shows highest homology to the Msh2 subgroup. msh2 disruption gives rise to increased mitotic mutation rates and increased levels of postmeiotic segregation of genetic markers. In bandshift assays performed with msh2Δ cell extracts, a general mismatch-binding activity is absent. By complementation assays, we showed that S. pombe msh2 is allelic with the previously identified swi8 and mut3 genes, which are involved in mating-type switching. The swi8-137 mutant has a mutation in the msh2 gene which causes a truncated Msh2 peptide lacking a putative DNA-binding domain. Cytological analysis revealed that during meiotic prophase of msh2-defective cells, chromosomal structures were frequently formed; such structures are rarely found in the wild type. Our data show that besides having a function in mismatch repair, S. pombe msh2 is required for correct termination of copy synthesis during mating-type switching as well as for proper organization of chromosomes during meiosis.

In Escherichia coli, the mutHLS system efficiently repairs single-base mismatches except C/C, as well as small single-strand insertions and deletions (45). In addition, this system maintains genome stability by prevention of recombination between homeologous sequences (55). The MutS protein recognizes and binds to mismatches. The site-specific endonuclease MutH binds to hemimethylated dam (GATC) sequences. MutL connects both complexes by binding to MutS and MutH. Upon complex formation, MutH is activated and initiates excision of the newly synthesized DNA strand, followed by resynthesis, resulting in intact duplex DNA (39, 45, 46, 47).

Several MutS and MutL homologs have been identified in eukaryotes (13, 39, 47), indicating that the system is more complex than in bacteria. In Saccharomyces cerevisiae, three MutS homologs, designated Msh2 (56), Msh3 (49), and Msh6 (34, 43), and two MutL homologs, Mlh1 and Pms1 (40, 54), have been shown to be involved in mismatch repair of nuclear DNA. Msh2 forms a complex with Msh3 to repair loops and with Msh6 to repair single-base mismatches (2, 31, 34, 43). msh2 mutants display increased mitotic and meiotic mutation rates (57) and instability of simple repeats (66). In addition, the major mismatch-binding activity of S. cerevisiae cell extracts is absent in msh2 mutants (44). The mechanism of mismatch repair involving heterodimers of MutS and MutL homologous proteins is conserved in higher eukaryotes (reviewed in references 36, 46, 47, and 70). Defects in the human mismatch repair system give rise to increased microsatellite instability and predisposition to a common form of colon cancer (hereditary nonpolyposis colon cancer) (46, 70). Inactivation of the murine MSH2 gene results in increased mutation rates, microsatellite instability, and cancer but has no effect on fertility (15). In contrast, MLH1-deficient mice and PMS2-deficient male mice are sterile. They show an arrest in meiosis I and abnormal chromosome synapsis, respectively (7, 16).

For the fission yeast Schizosaccharomyces pombe, at least two pathways of mismatch repair have been postulated based on two lines of evidence. Marker effects of G-to-C transversions observed in intragenic crosses indicate the existence of a major pathway able to repair most base mismatches except C/C and of a minor pathway able to correct C/C and other mismatches (62, 63). In band shift assays with S. pombe wild-type cell extracts, two mismatch-binding activities were identified (23). One activity binds to small loops and to most single-base mismatches but not C/C; the second activity binds to C/C and all other cytosine-containing mismatches. Due to its substrate specificity and excision tract length, the major pathway was proposed to be homologous to the E. coli mutHLS system. To differentiate the functions of the two pathways, we engaged in the isolation of genes homologous to mutL and mutS. The characterization of the S. pombe mutL homolog pms1⁺ was described elsewhere (60). S. pombe swi4⁺, the homolog of S. cerevisiae MSH3, was identified due to its involvement in mating-type (MT) switching (21). Here we report the isolation of the S. pombe msh2⁺ gene and its involvement in mismatch repair, MT switching, and meiotic chromosome organization.

MATERIALS AND METHODS

Strains and media.

All S. pombe strains were derived from the original wild-type strain introduced by Leupold (reviewed in reference 29) and are listed in Table 1. Standard media and general genetic methods were described by Gutz et al. (29). YEA (yeast extract agar) medium containing a limiting amount of adenine or MMA (minimal medium) with limiting adenine supplementation (5 mg/liter) was used to investigate the colony color of ade6 mutants. Adenine limitation allows ade6 mutants to grow, but due to the block in adenine synthesis they accumulate a red pigment (29). MMA medium used for fluctuation tests contained 200 mg of guanine per liter to inhibit background growth of adenine auxotrophs (14).

TABLE 1.

S. pombe strains used in this study

Strain	Genotype	Reference or source
968	h90	29
972	h−	29
34-1344	h− ade6-51	Strain collection, University of Bern
57-2254	h− his3-D1	10
JB6	h+/h− ade6-M210/ade6-M216	J. Bähler
E113	h90 ade6-M210 his2 swi8-113	R. Egela
E120	h90 ade6-M210 his2 swi8-120	R. Egela
E131	h90 ade6-M210 his2 swi8-131	R. Egela
E137	h90 ade6-M210 his2 swi8-137	R. Egela
HE443	h90 ade6-M210 his2 swi8-2	K. Ostermannb
HE447	h90 ade6-M210 his2 swi8-3	K. Ostermannb
HE449	h90 ade6-M210 his2 swi8-4	K. Ostermannb
HE460	h90 ade6-M210 his2 swi8-5	K. Ostermannb
HE461	h90 ade6-M210 his2 swi8-6	K. Ostermannb
HE468	h90 ade6-M210 his2 swi8-7	K. Ostermannb
HE471	h90 ade6-M210 his2 swi8-8	K. Ostermannb
HE474	h90 ade6-M210 his2 swi8-9	K. Ostermannb
HE475	h90 ade6-M210 his2 swi8-10	K. Ostermannb
HE480	h90 ade6-M210 his2 swi8-11	K. Ostermannb
MAB033	h− ade6-M210 his3-D1 leu1-32 ura4-aim ura4-D18	8
MAB054	h− ade6-M210 his3-D1 msh2::his3+ ura4-aim ura4-D18	8
Ru31	h90 ade6-M210 his7 swi8-137 ura4-D18	This studya
Ru32	h90 ade6-M210 mut3-25 ura4-D18	This studyc
Ru34	h90 ade6-M210 lys1-131 swi8-49	This study
Ru37	h90 ade6-M210 his3-D1 ura4-D18	This study
Ru39	h− his3-D1 msh2::his3+	This study
Ru108	h+ ade6-M210 his3-D1 msh2::his3+	This study
Ru109	h90 ade6-M210 his3-D1 msh2::his3+ ura4-D18	This study
Ru125	h− ade6-51 his3-D1 msh2::his3+	This study
Ru189	h− ade6-M216 his3-D1 msh2::his3+	This study
Ru195	h+ ade6-704 his3-D1 msh2::his3+ sup3-UGA,r36	This study
Ru196	h− ade6-704 his3-D1 msh2::his3+ sup3-UGA	This study
Ru198	h+/h− ade6-M210/ade6-M216 his3-D1/his3-D1 msh2::his3+/msh2::his3+	This study
Ru211	h+ ade6-M26 his3-D1 msh2::his3+	This study
Ru260	h90 ade6-M210 lys1-131 swi8-1	This studyb

^aAlleles swi8-49, swi8-113, swi8-120, swi8-131, and swi8-137 were isolated by R. Egel (Copenhagen, Denmark) after ethyl methanesulfonate mutagenesis. 

^bAlleles swi8-1 to swi8-11 were isolated by K. Ostermann (Braunschweig, Germany) in a two-step mutagenesis with UV irradiation and ethyl methanesulfonate treatment (22). 

^cThe mut3-25 mutant was isolated by P. Munz (Bern, Switzerland) after UV irradiation (48). 

PCR, gene cloning, and physical mapping.

To clone the S. pombe msh2⁺ gene, we performed PCR with degenerate primers derived from amino acid sequences highly conserved among MutS proteins. The 5′ oligonucleotide 5′-GCTCTAGACNGGNCCNAA(C/T)ATGGG-3′ was derived from the peptide motif TGPNMG (amino acids 687 to 692 in S. cerevisiae Msh2). The 3′ oligonucleotides 5′-CACGGTACCNCGNCC(C/T)AA(C/T)TC(A/G)TC-3′ and 5′-CACGGTACC(C/T)CTNCC(C/T)AA(C/T)TC(A/G)TC-3′ were derived from the consensus sequence DELGRG (amino acids 767 to 772 in S. cerevisiae Msh2). An XbaI site in the 5′ primer and KpnI sites in the 3′ primers were used for cloning of the resulting PCR fragments. PCRs were performed in 100 μl containing 250 to 500 ng of S. pombe genomic DNA, 200 pmol of each oligonucleotide primer, 0.1 U of Perfect Match Polymerase Enhancer (Stratagene), 1 U of Tfl DNA polymerase (Epicentre), 1× Tfl buffer, 1.5 mM MgCl₂, and 1.5% dimethyl sulfoxide. Reactions consisted of 35 cycles of 45 s at 94°C, 1 min at 45°C, and 1 min at 72°C. Fragments of about 270 bp in size were eluted from an agarose gel and reamplified by 25 cycles of 45 s at 94°C, 1 min at 42°C, and 1 min at 72°C. PCR products were digested with XbaI and KpnI, cloned into M13mp18, and sequenced with universal primers. Based on the sequence of Msh2-specific inserts, the primers Msh2-inv-up (GTGTGGCATGCCGGCAATAACTCCAAC) and Msh2-inv-down (CAGTGCCTTGTGAAGTGGCTGATATCTAG) were synthesized. For inverse PCR, genomic S. pombe DNA was digested with SpeI and religated with T4 DNA ligase (Boehringer Mannheim). Using the religated SpeI fragments as templates, PCRs with the oligonucleotides Msh2-inv-up and Msh2-inv-down were carried out. An SphI recognition site was introduced in Msh2-inv-up, an EcoRV recognition site was introduced in Msh2-inv-down. PCRs consisted of 35 cycles of 45 s at 94°C, 30 s at 60°C, and 3 min at 70°C, using the reaction conditions described above. Identity of the resulting 1.6-kb fragment was confirmed by sequencing the ends of cloned PCR products. A 0.8-kb SpeI/SphI fragment of the 1.6-kb inverse-PCR fragment was used to hybridize S. pombe cosmid and P1 phage libraries gridded on high-density filters (32). Eighteen positive clones were obtained, and all map to the chromosome II interval between cdc10 and his2.

Subcloning, nucleotide sequence analysis, and gene disruption.

Restriction mapping was performed by Southern analysis of digested cosmid clone c24C6 (Fig. 1A). A 4.3-kb SacI fragment and a 4-kb EcoRI/HindIII fragment were subcloned into pUC18, resulting in plasmids pRU20 and pRU21, respectively (Fig. 1A). A 6.3-kb HindIII fragment was cloned into pUR19 to give rise to pRU22. Nested deletions of pRU20 were obtained by exonuclease III treatment (Erase-a-Base system; Promega) and were sequenced by using either a universal or reverse M13 primer or synthesized oligonucleotides derived from internal sequences (data not shown). The sequence upstream of the internal SacI site was determined by using plasmid pRU21 and synthesized oligonucleotides. Sequencing reactions on both DNA strands of msh2⁺ were performed by the dideoxy-chain termination method, using a sequencing kit (United States Biochemical Sequenase kit). For sequencing the swi8-137 mutation, PCR was performed on genomic DNA of strain E137. A fragment of about 1.9 kb containing the 5′ part of msh2 was obtained with primers P26 (5′-TGGTGTTTAATAGTTCGAATGC-3′) and P5 (5′-CTCCTCATCAAACTCTGCACGG-3′), and an ∼1.7-kb fragment containing the 3′ part of msh2 was obtained with primers P4 (5′-CACTTCCGAAGATCGTTACGGT) and 3′msh2 (5′-CTTCCAAAAAACATGTACCTTGG-3′). For both reactions, the program consisted of an initial 5-min denaturation at 94°C, 30 cycles of 45 s 94°C, 45 s 48°C, and 2.5 min 72°C, and finally an extension step of 10 min at 72°C. The PCR products were separated from primers, and the single coding strands were amplified by 20 cycles with either primer P5 or primer 3′msh2. Sequences were determined with internal primers by the dideoxy method. From the region containing the swi8-137 mutation, the complementary strand was also amplified and sequenced.

FIG. 1.

Structure of the msh2⁺ gene and the msh2::his3⁺ insertion mutation. (A) Restriction fragments containing the msh2⁺ coding region. The 4.3-kb SacI fragment and the 4-kb HindIII/EcoRI fragment were subcloned from cosmid c24C6 (32) into pUC18, giving rise to plasmids pRU20 and pRU21, respectively, and were used for sequencing. The 6.3-kb HindIII fragment was inserted into vectors pUR19 and pBluescript KS, resulting in plasmids pRU22 and pRU23. pRU22 was used for complementation assays, while pRU23 was the basis for the msh2::his3⁺ gene disruption cassette. (B) Insertion of the his3⁺ marker at the msh2 locus. The open box represents the msh2⁺ coding region; dark grey boxes indicate the two introns. The his3⁺ marker (light grey box) was inserted into the MluI and StyI restriction sites of pRU23 as shown to give rise to plasmid pRU24. The resulting msh2::his3⁺ HindIII fragment was transformed into strain 57-2254. It was also subcloned into the vector pUR19, resulting in pRU28.

For gene disruption, the 6.3-kb HindIII fragment containing the msh2⁺ open reading frame (ORF) and flanking sequences was cloned into pBluescript KS, resulting in pRU23. MluI and StyI restriction sites at the 5′ and 3′ ends, respectively, of the msh2⁺ gene were used to integrate the his3⁺ marker (10). For this purpose, MluI and StyI sites at the 3′ and 5′ ends, respectively, of his3⁺ were introduced by PCR. By this insertion, 94% of the msh2⁺ ORF was deleted (Fig. 1B). From the resulting plasmid (pRU24), the msh2::his3⁺ fragment was isolated after digestion with PstI and SalI and used to transform the strain 57-2254. Transformation was carried out by the lithium acetate method (35). One transformant (Ru39) with the correct msh2::his3⁺ gene disruption, as identified by Southern analysis (data not shown), was stored and used for all further crosses and experiments.

The msh2::his3 fragment obtained by PstI and SalI digestion of pRU24 was also subcloned into the vector pUR19, resulting in plasmid pRU28. This plasmid served as control in the complementation assays.

Analysis of msh2⁺ expression.

A meiotic time course was performed with the diploid wild-type strain JB6 as described by Bähler et al. (6). Samples were taken before the shift to meiosis (mitotic probe) and every 2 h thereafter. Total RNA isolation and Northern analysis were performed as described elsewhere (27), using a 0.9-kb msh2⁺ fragment (data not shown) as the hybridization probe. Signals were quantified with a PhosphorImager.

Mutator phenotype and mutation rate determination.

S. pombe ade6 mutants were grown on YEA medium to visualize the red colony color. White sectors in red colonies and white colonies originate by forward mutations in genes upstream of ade6 in the adenine pathway, by mutations in suppressor genes, or less frequently by reversion of the original ade6 mutations. White-sectored colonies were rare in wild-type strains but occurred in the majority of colonies in msh2, swi8, and mut3 strains.

For determination of spontaneous mutation rates, single colonies from fresh plates were used to inoculate 3 ml yeast extract liquid cultures. After growth at 30°C for 36 h, appropriate dilutions were plated on YEA to determine the total viable cell number. The remainder of the cultures was pelleted; then either the cells were plated on MMA plates containing 200 mg of guanine per liter for the selection of ade6-51 revertants or appropriate dilutions were distributed on YEA plates containing 1 g of 5-fluoro-orotic acid per liter (26) to determine the frequency of ura4/ura5 mutants. To determine the frequency of Ade⁺ revertants, 15 cultures per experiment were used, and colonies were counted after 7 days of growth. To determine the frequency of Ura⁻ mutants, each experiment included seven cultures, and colonies were counted after 6 days of growth. Spontaneous mutation rates were determined by the method of the median (41). To determine the mutational spectra of msh2 and wild-type strains, Ura⁻ mutants derived from independent cultures were crossed with ura4 mutants to discriminate whether the mutation is located in ura4 or ura5 and, if it is located in ura4, to estimate the position of the mutated site in the ura4 gene. The mutated site was then determined by sequencing of PCR products as described elsewhere (5).

Tetrad analysis for determination of PMS.

Tetrad analysis was performed to determine the spore viability and the frequency of postmeiotic segregation (PMS) in the msh2 background. Two pairs of parental strains, both homozygous for msh2::his3⁺, were used. One pair was heterozygous for the ade6-M26 marker (ade6-M26 versus ade6⁺); the second pair was heterozygous for sup3 (sup3-UGA versus sup3-UGA,r36). sup3-UGA but not sup3-UGA,r36 suppresses the ade6-704 mutation (38). Tetrads were dissected on YEA and visually analyzed after 5 days of growth at 30°C. All spore clones from aberrant tetrads were investigated for adenine prototrophy, mating type, and ploidy to exclude tetrads unlikely to result from PMS or whole chromatid conversion (WCC) events. Only tetrads consisting of four viable spores were considered. To determine whether the ratio of PMS to WCC in different crosses was significantly changed, we used χ² tests with 2 × 2 tables. Standard deviations (SD) were calculated according to the formula SD = [p(100 − p)/n]^1/2, where p is the percentage of aberrant events and n is the total number of tetrads.

Mismatch binding assay.

Mismatch-binding specificities of wild-type and msh2 cell extracts were tested by a gel retardation assay as described by Fleck et al. (23). Approximately 50 μg of protein extracts was incubated with 40 fmol of radiolabeled oligonucleotides and with 1.6 pmol of unlabeled competitor DNA (homoduplex with the same sequence context) for 20 min at 4°C. Reactions were carried out in a mixture containing 25 mM Tris-HCl (pH 7.5), 0.5 mM dithiothreitol, 4 mM spermidine, 0.5 mM EDTA, 10% glycerol, 50 mM NaCl, 25 mM KCl, 0.01 mM ZnCl₂, 0.1 mM dATP, 0.1 mM dCTP, 0.1 mM dGTP, and 0.1 mM dTTP. Electrophoresis of reaction mixtures in nondenaturing gels was performed at 120 V and 4°C in 40 mM Tris-HCl (pH 7.5)–0.4 M sodium acetate–0.5 mM EDTA. Substrates and competitor DNA used for the assay are derived from the M13mp9 sequence (23).

Iodine staining.

S. pombe strains were grown on sporulation medium (MEA) for 3 days and treated with iodine vapor. The procedure stains the spores, resulting in brown colonies (29). Homothallic (h⁹⁰) wild-type strains exhibit homogeneous staining, referred to as an iodine-positive phenotype. After iodine treatment, colonies of h⁹⁰ mutants partially defective in MT switching show iodine-negative sectors in the colonies, or a mottled phenotype (28). swi8, msh2, and mut3 mutants give rise to mottled colonies, and they frequently segregate iodine-negative, nonsporulating colonies. This phenotype is specific for mutants of class II switching genes of S. pombe, which are thought to be defective in the termination step of copy synthesis during MT switching (17, 20, 22, 30).

Complementation assays.

S. pombe strains were transformed by the lithium acetate method (35) and plated onto MMA. After 7 days, transformants were replica plated onto YEA and MEA to examine complementation of the mutator phenotype and of the switching defect, respectively. For each transformation, 100 colonies were analyzed by careful visual examination. For complementation of the MT switching defect, only iodine-positive or mottled colonies, not iodine-negative colonies, were considered. The latter are due to rearrangements in the MT region which cannot be complemented (22).

Cytological procedures.

To analyze the formation of linear elements in meiotic prophase, time course experiments were performed with the diploid msh2Δ strain Ru198 and with the diploid wild-type strain JB6. Ru198 was constructed from strains Ru108 and Ru189 by intragenic complementation of the alleles ade6-M210 and ade6-M216 (29). Time course experiments were performed as described by Bähler et al. (6), and samples were taken before and at hourly intervals after the shift to meiosis. To investigate the formation of linear elements, each sample was spread, silver stained, and analyzed by electron microscopy as described elsewhere (6). For each time point, 100 to 200 nuclei were examined. To monitor successful progression of meiosis, samples were treated with the DNA-staining dye 4′,6-diamidino-2-phenylindole (DAPI), and the percentage of cells containing more than one nucleus was determined by fluorescence microscopy. At least 150 cells were analyzed for each time point.

Nucleotide sequence accession number.

The S. pombe msh2 sequence has been deposited in the EMBL database under accession no. AJ006948.

RESULTS

Cloning and sequencing of msh2.

Identification of the S. pombe msh2 gene was based on PCR with degenerate primers as described in Materials and Methods. Screening of S. pombe cosmid and P1 phage libraries (32) with an msh2-specific probe allowed the identification of 18 clones which all mapped to the interval between cdc10 and his2 on chromosome II. Sequence analysis revealed an ORF of 2,949 bp, interrupted by two potential introns of 43 and 94 bp at the 5′ end (Fig. 1). PCR using a cDNA library (18) as the template and sequencing of the PCR fragments cloned into the vector pGEM3 confirmed the existence of both introns. Both contain 5′ and 3′ splice sites as well as branch sites according to the consensus sequences of Prabhala et al. (53). Intron I is located between bp 32 and 74 (with respect to the start codon) with the 5′ and 3′ splice sites GTTTGT and CAG, respectively, and the branch site CTAAT. Intron II reaches from bp 158 to 251 with the 5′ and 3′ splice sites GTAAGT and TAG, respectively, and the branch site CTAAC. The translated msh2 ORF product consists of 982 amino acid residues which show 43% identity to S. cerevisiae Msh2 and human MSH2 and 28% identity to E. coli MutS. Amino acid sequence comparison of MutS homologs revealed that the new S. pombe MutS homolog belongs to the MSH2 subfamily.

To monitor expression of S. pombe msh2 in meiosis, we performed a meiotic time course with the wild-type strain JB6. Northern analysis with RNA isolated before (mitotic cells) and at different time points after the shift to meiosis revealed two transcripts, of 2.5 and 3.2 kb (data not shown). The length of the 3.2-kb transcript corresponds well to the expected size of mRNA, while the shorter transcript might be due to an internal transcription initiation site. The level of mRNA in mitotic cells was slightly higher than the constant amount during meiosis (data not shown).

Disruption of the msh2 gene increases spontaneous mitotic mutation rates.

An msh2::his3⁺ fragment derived from plasmid pRU24 (Fig. 1A) was used to transform strain 57-2254. For the resulting strain Ru39, correct integration of the his3⁺ marker into the msh2⁺ ORF was confirmed by Southern analysis (data not shown). To monitor spontaneous mitotic mutation rates in the msh2Δ mutant, we used two different systems. Rates of reversion from Ade⁻ to Ade⁺ were determined by using the ade6-51 mutation, a C-to-T transition (62). Ade⁺ mutants can arise due to forward mutations in suppressor genes or by reversion of the original ade6-51 mutation. In this assay, the spontaneous mitotic mutation rate was 15-fold higher in the msh2Δ strain Ru125 than in the wild-type strain 34-1344 (Table 2). The forward mutation rates at ura4⁺ and ura5⁺ were measured for strains Ru39 and 972 as well as strains MAB033 and MAB054. In the latter two strains, the ura4 gene is inserted about 15 kb upstream of the ade6 gene (25). Table 2 shows that the spontaneous mitotic mutation rate for ura4⁺ and ura5⁺ was increased 14- to 15-fold in the msh2Δ mutant. We determined the nature of six mutations inactivating ura4 in both wild-type and msh2Δ mutant backgrounds. In the wild-type background, five of six mutations were base substitutions (two C to T, two G to A, and one G to T); the sixth mutation was a deletion of an A at a site of two A’s (A₂ to A₁). In the msh2Δ background, three base substitutions (two G to T and one A to G) and three one-nucleotide deletions (two T₅ to T₄ and one A₅ to A₄) were determined. These data indicate that deletion of one nucleotide in a homonucleotide run frequently occurs in msh2-defective cells. These deletions are likely to result from DNA strand slippage. Nevertheless, base substitutions which originated from unrepaired base-base mismatches are also accumulated in msh2Δ cells.

TABLE 2.

Spontaneous mitotic mutation rates

Relevant genotype	Strain	Mean forward mutation rate from Ura+ to Ura− ± SDa	Mean mutation rate from ade6-51 to Ade+ ± SD	Fold increase
msh2+	972	(3.7 ± 2.3) × 10−8
msh2Δ	Ru39	(5.2 ± 0.45) × 10−7		14
msh2+	MAB033b	(4.9 ± 2.0) × 10−8
msh2Δ	MAB054b	(7.4 ± 4.5) × 10−7		15
msh2+	34-1344		(3.7 ± 0.14) × 10−10
msh2Δ	Ru125		(5.4 ± 0.99) × 10−9	15

^aSD = [Σ(x − x*)²/n − 1]^1/2, where x is the mutation rate from one experiment, x* is the mean value, and n is the number of experiments (=2). Fluctuation tests were performed as described in Materials and Methods. 

^bThe ura4 gene is inserted about 15 kb upstream of the ade6 gene (25). The original ura4 gene is deleted (ura4-D18). 

Increase of PMS and reduction of spore viability.

During meiotic recombination, heteroduplex DNA is formed. When mismatch repair fails for a given marker, PMS tetrads are observed. They include 5⁺:3⁻, 3⁺:5⁻, or aberrant 4⁺:4⁻ segregations resulting from persistence of mismatches in spores. Germination of such spores gives sectored colonies. In contrast, 6⁺:2⁻ or 2⁺:6⁻ gene conversions may originate from repair of mismatches in heteroduplex DNA. To investigate the ability of the msh2Δ mutant to repair mismatches arising during meiotic recombination, we performed tetrad analysis and determined the PMS and gene conversion frequencies with two pairs of strains. Ru195 and Ru196 are homozygous for msh2::his3⁺ and heterozygous for the functional (sup3-UGA) and nonfunctional (sup3-UGA,r36) suppressor of the ade6-704 mutation (38). Formation of heteroduplex at sup3 gives rise to G/T and A/C mismatches. The PMS frequency in this cross was increased from 0.1% in the msh2⁺ cross to 1.9% in the msh2Δ mutant, while the gene conversion frequency was reduced from 1.5% to 0.2% (Table 3). Therefore, the percentage of PMS events among all aberrant tetrads raised from 5.9% in the msh2⁺ cross to 91% in the msh2Δ mutant. The second cross for determination of PMS and conversion frequencies was performed with the msh2Δ strains Ru39 (ade6⁺) and Ru211 (ade6-M26). Heteroduplex DNA formed at ade6-M26 (52, 69) results in G/A and T/C mismatches. In this cross, the PMS frequency was shifted from 0% in the wild type to 5.2% in the msh2Δ background, while the conversion frequency was reduced from 5.1% to 0.9%. The PMS frequency among all aberrant tetrads increased from less than 1.9% in the wild type to 85% in the msh2Δ mutant (Table 3). The increase in PMS frequency was highly significant for both crosses (χ² = 20.2 for sup3-UGA,r36; χ² = 62.5 for ade6-M26; χ²_P0.01 = 6.635).

TABLE 3.

PMS and gene conversion frequencies in msh2Δ tetrads

Relevant genotype	Conversion			PMS				Total no. of tetrads	Mean % ASa ± SDb	% PMSc
	6+:2−	2+:6−	%	5+:3−	3+:5−	ab4+:4−d	%
sup3-UGA/sup3-UGA,r36
msh2+/msh2+e	7	9	1.5	0	1	0	0.1	1,105	1.5 ± 0.37	5.9
msh2Δ/msh2Δ	0	1	0.2	5	5	0	1.9	514	2.1 ± 0.63	91
ade6+/ade6-M26
msh2+/msh2+f	46	6	5.1	0	0	0	0	1,018	5.1 ± 0.69	<1.9
msh2Δ/msh2Δ	3	1	0.9	15	2	3	5.2	445	6.1 ± 1.13	85

^aAS, aberrant segregations. 

^bSD = [p(100 − p)/n]^1/2, where p is the percentage of aberrant segregations and n is the total number of tetrads analyzed. 

^cCalculated as PMS events (5⁺:3⁻ + 3⁺:5⁻ + 2 × ab4⁺:4⁻)/total number of aberrant segregations. 

^dAberrant 4⁺:4⁻ tetrads were counted as two PMS events. 

^eData from Schär et al. (60). 

^fData from Gutz et al. (29). 

In addition, we determined the spore viability in both crosses. Among 3,040 spores dissected from the cross of strains Ru195 and Ru196, 2,638 (87%) were viable. Crossing strains Ru39 and Ru211 gave rise to 2,301 viable spores out of 2,684 dissected spores (86%). Thus, the overall spore lethality was about 14% in the msh2Δ background. For wild-type strains, an overall lethality of 3 to 8% was observed (60), indicating a two- to threefold increase of spore lethality in the msh2Δ mutant.

The MutS-type mismatch-binding activity is absent in swi8-137 and msh2Δ cell extracts.

In gel retardation assays performed with S. pombe wild-type cell extracts, two mismatch-binding activities were discovered (23). One low-mobility complex binds efficiently to T/G mismatches and with various efficiencies to 1-bp deletions and most single-base mismatches but not to C/C and poorly to C/A. The second, high-mobility complex binds to C/C and all other kinds of cytosine-containing mismatches.

During the screen for mutants which lack one of the mismatch-binding activities, we also tested the swi8-137 mutant E137, which is defective in MT switching and shows a mutator phenotype (see below). As a control, we used wild-type cell extracts derived from strain 968. The wild type exhibits an activity of low mobility which specifically binds to T/G, but not to C/C or C/A, mismatches (Fig. 2, lanes 1 to 3) and therefore displays the same binding abilities as described previously (23). In contrast, this mismatch-specific activity is absent in protein extracts derived from the swi8-137 mutant (Fig. 2, lanes 4 to 6). The C/C-binding complex is present in wild-type and swi8-137 cell extracts. These data suggest that the low-mobility mismatch-binding activity but not the C/C-binding activity is dependent on Swi8.

FIG. 2.

Mismatch-binding activities of S. pombe wild-type (lanes 1 to 3) and swi8-137 (lanes 4 to 6) cell extracts. Substrates are T/G (lanes 1 and 4), C/C (lanes 2 and 5), and C/A (lanes 3 and 6). All reactions included a 40-fold excess of unlabeled T:A competitor. The upper arrow indicates the general mismatch-binding activity; the lower arrow indicates the C/C-binding activity.

We also performed bandshift assays using the msh2Δ strain Ru39. With this strain, we obtained the same results as with the swi8-137 mutant, indicating that the low-mobility complex is absent whereas binding of cytosine-containing mismatches by the high-mobility activity is unaffected (data not shown). Therefore, the low-mobility mismatch-binding complex but not the C/C-binding activity is dependent on Msh2.

The msh2 gene is allelic with swi8 and is involved in MT switching.

msh2 maps on chromosome II between cdc10 and his2. The swi8 gene, which is involved in MT switching, had been mapped to the same interval (30, 37). In swi8 mutants, duplications and deletions arise in the MT region due to failure of correct termination of copy synthesis during MT switching (22). In an ade6 background, swi8 mutant colonies display a mutator phenotype (Materials and Methods), indicating a possible function of swi8 in DNA repair (22). In addition the low-mobility mismatch-binding complex is absent in the swi8-137 mutant (Fig. 2).

We supposed that swi8 might be allelic with msh2. To test this possibility, we first examined whether msh2Δ mutant colonies show a mutator phenotype and defects in MT switching. Both phenotypes were apparent in strain Ru109. Complementation of the swi8 defects by msh2⁺ on a plasmid was then tested. For this purpose, an swi8-137 strain (Ru31), an msh2Δ strain (Ru109), and a wild-type strain (Ru37) were transformed with a plasmid harboring the complete ORF of msh2⁺ (pRU22 [Fig. 1A]). As controls, the same strains were transformed with the vector (pUR19) and with a plasmid containing the flanking regions of msh2 (pRU28). In this plasmid, the msh2::his3⁺ gene disruption cassette is inserted in the shuttle vector pUR19 (Fig. 1B). Transformants were replica plated onto YEA to examine the mutator phenotype and onto MEA to investigate the efficiency of MT switching. Complementation of the mutator phenotype was assayed by determination of the percentage of homogeneously red colonies versus white-sectored colonies. Transformation with msh2⁺ (pRU22) resulted in 77% red colonies in the swi8-137 mutant and 82% red colonies in the msh2Δ mutant (Table 4). In contrast, more than 90% sectored colonies were observed for the control transformations with the swi8-137 and the msh2Δ strains. In wild-type background, about 90% of red colonies were observed in all three transformations. These data indicate that the msh2⁺ gene located on plasmid pRU22 is functional and that pRU22 but none of the controls is able to complement the mutator phenotype of swi8-137 and msh2Δ mutants.

TABLE 4.

Complementation of the mutator phenotype

Plasmid	Colony type (%)b
	msh2Δa (Ru109)		swi8-137 (Ru31)		Wild type (Ru37)
	Red	White sectors	Red	White sectors	Red	White sectors
pmsh2+	82	18	77	23	88	12
pmsh2Δ	1	99	1	99	90	10
pUR19	3	97	6	94	92	8

^aFull genotypes are given in Table 1. 

^b100 colonies per transformation were analyzed. 

Complementation of the defects in MT switching was tested by determination of the percentage of iodine-positive colonies as described in Materials and Methods. The iodine-negative colonies frequently arising in swi8 (22) and msh2Δ mutants were not considered. Table 5 shows that msh2⁺ complements the defects of swi8-137 and msh2Δ mutants, while no complementation was found for the control transformations. All wild-type colonies analyzed were iodine positive. No complementation was found with plasmid pRU28, which contains the flanking regions of msh2⁺. Thus, complementation requires the msh2⁺ gene and not any other genes which might be located on plasmid pRU22.

TABLE 5.

Complementation of defects in MT switching

Plasmid	Iodine reaction (%)b
	msh2Δa (Ru109)		swi8-137 (Ru31)		Wild type (Ru37)
	Positive	Mottled	Positive	Mottled	Positive	Mottled
pmsh2+	95	5	96	4	100	0
pmsh2Δ	0	100	0	100	100	0
pUR19	0	100	0	100	100	0

^aFull genotypes are given in Table 1. 

^b100 colonies per transformation were analyzed. 

Sequencing of the msh2 gene of the swi8-137 mutant E137 revealed a mutation in the C-terminal part changing the CGA codon for arginine (R794) at nucleotide 2796 to a TGA stop codon (the ATG start codon is at position 280 of the sequence deposited to the EMBL database). This mutation produces a truncated Msh2 peptide without a potential DNA-binding domain (helix-turn-helix motif) and sequences highly conserved among MutS homologs (21, 49).

Genetic crosses indicated that swi8 is allelic with mut3 (22). The mut3-25 mutant was isolated after methyl methanesulfonate mutagenesis and screening for mutants showing a mutator phenotype (48). Using the same assays as for the swi8-137 mutant, we found that msh2⁺ is able to complement the defects of the mut3-25 mutant Ru32 (data not shown). Therefore, msh2 is also allelic with mut3.

Our data show that S. pombe msh2 is involved in MT switching and in mismatch repair. Both functions might be carried out by the same enzymatic activity (most likely recognition of mismatches or DNA loops), or they might be directed by different domains within the Msh2 protein. Therefore, we tested the 16 known swi8 alleles (Table 1) for a possible separation of function. All were isolated on the basis of defects in MT switching. We looked for swi8 mutants displaying no mutator phenotype. No such mutant was found. In addition, the mut3-25 allele, isolated due to its mutator phenotype (48), causes defects in MT switching (22). Therefore, no separation of function was observed among 17 msh2 alleles.

Altered linear element formation during meiotic prophase.

Involvement of mismatch repair proteins in basic mechanisms of meiosis, besides mismatch repair, has been reported for bacteria as well as for eukaryotes (reviewed in references 47 and 61). In the great majority of higher eukaryotes, a synaptonemal complex, consisting of two lateral elements (also called axial elements before formation of the complex) and a central element, is formed during meiotic prophase. In S. pombe, meiotic chromosome pairing and recombination do not require a synaptonemal complex. Instead, linear elements, probably corresponding to the axial elements of other eukaryotes, have been observed during meiotic prophase (reference 6 and references cited therein). In wild-type nuclei, different stages of prophase were defined by the characteristic organization of linear elements (6). Class I nuclei contain short elements and occur early in meiotic prophase. Later, the elements can form networks or bundles (class II nuclei). More abundant than class II are nuclei with long and separated filaments (class III nuclei). At the end of meiotic prophase, the elements seem to disintegrate to give rise to class I nuclei again (6).

To investigate whether S. pombe Msh2 is involved in meiotic prophase, we sporulated the diploid strain Ru198, homozygous for the msh2 deletion, and as a control the diploid wild-type strain JB6. Samples were taken shortly before shift to meiotic conditions (time point zero) and at every hour after the shift. They were analyzed for completion of the first meiotic division by DAPI staining, while the formation of linear elements was monitored by spreading, silver staining, and electron microscopy of nuclei, as described by Bähler et al. (6). We observed a strong increase in the frequency of class II nuclei (Fig. 3) in the msh2Δ mutant, while in the wild type this class was rare: the maximum frequency of class II nuclei in the wild type was 6% (Fig. 3A, 6 h), whereas it was more than 50% in the msh2Δ mutant (Fig. 3B, 5 h). A concomitant decrease of class III nuclei was observed in the msh2Δ strain: maximum values of 39% (Fig. 3A, 6 h) in the wild type and 16% (Fig. 3B, 7 h) in the msh2Δ strain were observed. Examples of a wild-type class III nucleus and of a typical class II nucleus from the msh2 mutant are shown in Fig. 4. Completion of the first meiotic division, as indicated by the amount of cells containing more than one nucleus in DAPI stains, was the same in the two strains (Fig. 3).

FIG. 3.

Linear elements in meiotic prophase. Meiotic time courses and preparations of samples were performed with diploid wild-type strain JB6 (A) and diploid msh2Δ mutant Ru198 (B). After spreading and silver staining of cells from each time point, 100 to 200 nuclei were analyzed by electron microscopy to determine the percentage of each class of nuclei. To determine the percentage of cells that had completed the first meiotic division (more than one nucleus per cell), at least 150 DAPI-stained cells were analyzed for each time point.

FIG. 4.

(A) Example of a class II nucleus typical for the msh2Δ mutant. The linear elements are aggregated and form densely stained networks and bundles. This class is abundant in the msh2Δ mutant but rare in the wild type (see text and Fig. 3). (B) Example of class III nucleus of the wild type. Linear elements are long and separated. Class III nuclei are found frequently in wild-type cells but less often in the msh2Δ mutant (see text and Fig. 3). Bars represent 1 μm.

DISCUSSION

Based on genetic data, at least two mismatch repair pathways were proposed for S. pombe (62, 63). Due to its substrate specificity and repair tract length, the major pathway was suggested to be homologous to the E. coli mutHLS system. To study the features of the two pathways in more detail, we engaged in the identification of genes that are involved in mismatch repair in S. pombe. The properties of the mutL homolog pms1 were described recently (60). Here we report the cloning and characterization of msh2, a mutS homolog of S. pombe. Among the known MutS homologous proteins, Msh2 shows highest homology to the Msh2 subgroup. This and the phenotypic characterization of the msh2Δ strain revealed that we have identified an important component of the major mismatch repair pathway.

The msh2 gene is involved in mitotic and meiotic mismatch repair.

To investigate the function of msh2⁺ in postreplicative mismatch repair, we determined mitotic mutation rates in msh2Δ mutants (Table 2). We found a 15-fold increase of the reversion rate of the ade6-51 marker (a C-to-T transition). Reversions of ade6-51 likely occur by nonrepaired base-base mismatches. The spontaneous mutation rate at ura4⁺/ura5⁺ was increased 14- to 15-fold. To explore the mutational spectra, we determined several ura4 inactivating mutations from wild-type and msh2Δ strains. In the wild-type background, we found two C to T, two G to A, one G to T, and one A₂ to A₁. Thus, five of six analyzed mutations were base-base substitutions, while the sixth was a deletion of one A at a site of two A’s. In the msh2Δ background, we found two G to T, one A to G, two T₅ to T₄, and one A₅ to A₄. Thus, 50% of the mutations were one-nucleotide deletions within mononucleotide runs, a type of mutation not observed in the wild type. The mutational spectra indicate that one-nucleotide loops within repeated mononucleotides occur frequently by DNA strand slippage and that Msh2 is required for their repair. Taken together, our data show that Msh2 is involved in repair of base-base mismatches as well as of single unpaired nucleotides which arise during the mitotic cell cycle. Similar results were reported for Msh2 of S. cerevisiae (43).

Current models of meiotic recombination include the formation of heteroduplex DNA. Mismatches arise when a heteroduplex is formed between two homologous but not completely identical sequences. Mismatch repair can lead to either restoration or gene conversion, while nonrepaired mismatches give rise to PMS events. Therefore, the frequency of PMS tetrads among all aberrant tetrads is indicative of the efficiency of mismatch repair during meiosis. The frequency of all aberrant tetrads reflects the frequency of heteroduplex formation (3, 51). We used two types of crosses to determine PMS and gene conversion frequencies in the msh2Δ mutant. In the cross sup3-UGA × sup3-UGA,r36, the mispairs G/T and A/C can be formed; in the cross ade6⁺ × ade6-M26, the mispairs G/A and T/C can arise. In both crosses, the PMS frequencies were significantly higher in the msh2Δ mutant than in the wild type (Table 3). Consistently, strongly reduced conversion frequencies were measured. Thus, the msh2Δ mutant is deficient in repairing mismatches in heteroduplex DNA. Interestingly, the total frequency of aberrant tetrads was not significantly altered in the msh2Δ mutant (Table 3). If in the wild-type background, heteroduplex DNA was repaired randomly toward wild-type or mutant information, and one half of the repair events would result in undetectable 4⁺:4⁻ restorations. In this case, inactivation of the mismatch repair system should increase the frequency of aberrant tetrads about twice since one half of the PMS events should then result from mismatches which if repaired would lead to restorations. In contrast, we found similar frequencies of aberrant tetrads for the msh2Δ mutant and for the wild type. This finding is consistent with the data reported by Schär et al. (60), who investigated PMS and gene conversion frequencies in S. pombe pms1 mutants, and of Alani et al. (3), who studied the effects of mutated msh2 of S. cerevisiae. It is concluded that mismatches in heteroduplex DNA are preferentially repaired toward gene conversion.

The msh2 gene is involved in the major pathway of mismatch repair.

The data for meiotic PMS frequencies and mitotic mutation rates show that the msh2⁺ function is necessary for efficient repair of mismatches arising in meiosis and during the mitotic cell cycle. The observed increase in PMS frequencies and in mitotic mutation rates caused by msh2Δ is comparable to the results obtained for the S. pombe pms1Δ mutant (60). Formation and repair of C/C and G/G mismatches in S. pombe meiosis was studied according to the protocol described by Lichten et al. (42). Spore DNA isolated from wild-type strains showed no G/G mismatches, while C/C mismatches were detectable. In contrast, G/G mismatches were detected in spore DNA isolated from msh2Δ or pms1Δ mutants, while the level of C/C mismatches was comparable to the wild-type level (8). Schär et al. (60) described for the pms1Δ mutant a significant increase in PMS frequency when they used marker combinations giving rise to G/T and A/C or to T/C and G/A mismatches. In contrast, in crosses giving rise to C/C and G/G mismatches, the increase in PMS frequency was strikingly low, obviously because C/C repair is not carried out by the major system. These results suggest that msh2⁺ and pms1⁺ (60) are involved in the major but not in the minor pathway of mismatch repair in S. pombe.

Biochemical indication for involvement of msh2 (swi8) in the major pathway of mismatch repair was obtained by bandshift assays performed with swi8-137 and msh2Δ mutants. In wild-type cell extracts of S. pombe, two mismatch-binding activities were identified (23). One of them was concluded to be part of the major mutLS-like repair pathway. This binding activity, being able to bind to small insertions and deletions and to most single-base mismatches, but not to C/C, is absent in the swi8-137 mutant. As swi8 is allelic with msh2, we also tested the msh2Δ mutant for its mismatch-binding abilities. Like in the swi8-137 mutant, the low-mobility activity thought to be part of the mutLS-like pathway is absent in msh2Δ cell extracts. In contrast, the mismatch-binding activity is not affected in cell extracts of a swi4Δ strain (23). Thus, like in other organisms (2, 24, 31, 34), Msh2, probably in a complex with Msh6, binds to base-base mismatches and small loops, while Swi4 in a complex with Msh2 binds to loops but not to base-base mismatches. The second activity, specifically binding to cytosine-containing mismatches including C/C, is still present in msh2Δ extracts. As the second activity was inferred to be part of the minor mismatch repair pathway, we conclude also from our biochemical data that msh2⁺ is involved in the major but not the minor pathway of mismatch correction.

Involvement of msh2 in MT switching.

swi8 mutants have been isolated according to their defects in MT switching. In homothallic strains, these defects cause a mottled colony phenotype and frequent segregation of iodine-negative colonies (17, 30). In addition, swi8 mutants have a mutator phenotype, indicating an increased spontaneous mutation rate (22). We found the same phenotypes for the msh2Δ mutant. Evidence that msh2 is allelic with swi8 (Table 4 and 5) and also with the mutator gene mut3 (48) was obtained from complementation assays. In these experiments, the defects in MT switching and the mutator phenotype of the swi8-137 mutant and of the mut3-25 mutant were complemented by msh2⁺. No complementation was observed when the same strains were transformed with the vector or with a construct harboring only the flanking regions of msh2⁺. These findings prompted us to conclude that complementation of the swi8-137 and mut3-25 defects is specifically dependent on msh2⁺. That msh2 is indeed allelic with swi8 was proved by sequencing the msh2 gene of the swi8-137 mutant E137. We found a mutation near the 3′ end of the gene which creates a TGA stop codon. The truncated Msh2-137 peptide lacks the C-terminal 189 amino acids. This region contains a potential DNA-binding domain (helix-turn-helix motif) and sequences highly conserved among MutS homologs (21, 49).

Like msh2⁺, swi4⁺, the S. pombe homolog of MSH3, is involved in MT switching (17, 20, 21, 30). It is known from other organisms that Msh2 forms a complex with Msh3, which binds to DNA loops (1, 31, 43). MT switching in S. pombe is initiated at a double-strand break at the right end of mat1 (9, 17, 50). Subsequently, copy synthesis using one of the silent cassettes (either mat2 or mat3) as the template and termination at the left-flanking homology boxes H2 and H3 is thought to occur (17, 20). swi4 and swi8 belong to the class II of swi genes which are thought to be defective in termination of copy synthesis (17, 20–22). Inactivation of swi4 or swi8 results in frequent duplications in the MT region. It was proposed that the duplications arise when copy synthesis is not terminated in the H2 and H3 homology boxes. In consequence, intervening sequences and the next MT cassette are synthesized (17, 20). An intramolecular secondary structure in the H2 or H3 homology box which might serve as a signal for correct termination was proposed (reference 21 and Fig. 5). This stem-loop like structure containing single-stranded loops might be bound by the MutS homologous proteins Msh2 and Swi4. Also, Swi10 and Rad16, the homologs of the nucleotide excision repair proteins Rad10 and Rad1 of S. cerevisiae, respectively, have a function in the termination step of MT switching (11, 58). It was recently reported that S. cerevisiae Rad1/Rad10 and Msh2/Msh3 act in the same pathway of mitotic recombination (59, 67). It was proposed that Msh2 and Msh3 of S. cerevisiae bind to branched DNA structures with free 3′ tails to enable removal of nonhomologous DNA by the Rad1/Rad10 endonuclease (67). In contrast, the MutL homologous proteins Pms1 and Mlh1 have no role in this pathway. Similar functions might be required for the MT switching process of S. pombe. The Rad16-Swi10 endonuclease and the Msh2-Swi4 complex may act in concert on the proposed secondary structure to ensure proper resolution of a specific recombination intermediate. Msh2 and Swi4 might bind to DNA loops and/or, as proposed for S. cerevisiae Msh2 and Msh3, to branched DNA. Both types of DNA structures are present when the secondary structure is formed (Fig. 5). The MutL homolog Pms1 is not involved in MT switching of S. pombe (19). We cannot rule out the possibility that another, not yet identified MutL homolog in S. pombe has a function in MT switching. However, analogous to the functions of respective proteins in S. cerevisiae, we propose that correct termination of MT switching in S. pombe requires binding (by Msh2/Swi4) but not repair (mediated by MutL homologous proteins) of loop-containing and/or branched DNA structures. Binding by Msh2/Swi4 might facilitate correct termination of DNA synthesis and resolution of the intermediate carried out by the endonucleolytic activity of Rad16/Swi10.

FIG. 5.

Hypothetical intramolecular secondary DNA structure in the H3 and H2 homology boxes of the silent cassette mat2. This structure or a part of it may be formed during MT switching. Binding of the MutS homologous proteins Msh2 and Swi4 to the DNA structure may terminate copy synthesis at the template DNA strand. Only the secondary structure of the template strand is shown. A similar structure is possible in the H3 and H2 homology boxes of mat3. The 5′ ends of the H3 and H2 boxes are indicated.

In swi8 mutants, a specific conversion event near the H1 box of mat1 was also observed (22). The conversion may occur when heteroduplex DNA formed during MT switching is extended into the region adjacent to the H1 homology box of the silent cassette mat2. This region shows some similarity to the corresponding region near the expression cassette mat1. In wild-type strains the conversion event is accurately suppressed by the swi8⁺ gene product. This function is thought to be similar to the antirecombination mechanism, which prevents recombination between similar but diverged sequences (22, 55). Our data strongly support this hypothesis: swi8 was shown to be allelic with the mutS homolog msh2, which in other organisms is known to prevent interspecies and homeologous recombination (12, 33, 55, 65).

Role of Msh2 in early meiotic prophase.

During meiosis, the homologous chromosomes pair and recombine. Not much is known about the molecular mechanisms of homolog recognition, a prerequisite for proper homologous pairing and recombination. It has been demonstrated in both bacteria and eukaryotes that different mismatch repair proteins play a role in the prevention of homeologous or ectopic recombination (12, 33, 55, 65). In both yeast and mammalian cells, the suppression of homeologous recombination is Msh2 dependent (reviewed in references 47 and 61). In an S. pombe meiotic time course with kinetics similar to those presented in this study, stable meiotic heteroduplex DNA is detectable shortly before the first meiotic division, around 7 h after meiotic induction (8). Compared to the wild type, the msh2Δ mutant shows a strong increase in the amount of class II nuclei with aggregated linear elements at the expense of class III nuclei with separated linear elements. As shown in Fig. 3, the frequency of class II elements in the msh2 mutant is higher than in the wild type after 3 h and increases to more than 50% after 5 h. Assuming that Msh2 acts on mismatches in heteroduplex DNA, this finding suggests that the protein has a role earlier in meiotic prophase before the formation of stable heteroduplex DNA, presumably by interacting with unstable heteroduplex DNA. As this heteroduplex DNA was not detected by Baur et al. (8), it might be below the detection limit of the assay. We can only speculate about the nature of this unstable heteroduplex DNA in S. pombe meiosis. More is known about the progression of meiotic recombination in S. cerevisiae: meiotic double-strand breaks appear and are resected early in meiotic prophase, before synaptonemal complex formation (68). They disappear concomitantly with the formation of double Holliday junctions, coinciding with formation of the synaptonemal complex. Double Holliday junctions persist throughout meiotic prophase and disappear when stable heteroduplex DNA is detectable and crossover products are formed (reference 68 and references cited therein). Swacha and Kleckner (68) presented a model in which mismatched base pairs are formed (and repaired) at two different time points during meiotic prophase: the first time early in meiosis, when double Holliday junctions are formed, and a second time at the end of meiosis, when the double Holliday junctions are resolved. It is likely that Msh2 binds to mismatches in heteroduplex DNA which is formed shortly before meiosis I at the end of meiotic prophase, because in tetrad analysis we observed strongly reduced repair of mismatches (Table 3). We propose that Msh2 also binds to heteroduplex DNA that is formed shortly after the initiation of meiotic recombination early in meiotic prophase, since the msh2Δ mutant shows an effect early in meiosis (Fig. 3 and 4). Probably S. pombe Msh2 has a role in the prevention of ectopic chromosome pairing. Msh2 binding may result in the rejection of heteroduplex DNA early in meiotic prophase, when the similarity or length of paired sequences is limited. A lack of Msh2 would then lead to an increase of ectopic interactions. Class II nuclei might correspond to a stage in meiotic prophase where the homology search takes place, reflected by an increase of ectopic interactions. The increased number of class II nuclei in the msh2Δ mutant may indicate increased and/or temporally extended pairing between homeologous sequences on nonhomologous chromosomal regions which results in an increase of ectopic interactions. This is the first indication of a possible biological role of the aggregation of linear elements in the class II nuclei and of a role of Msh2 in the homology search. It will be interesting to study the effects of msh2Δ on the frequency of ectopic pairing and recombination between identical and similar DNA sequences on different chromosomes in fission yeast meiosis (4, 64).