Swi5 Acts in Meiotic DNA Joint Molecule Formation in Schizosaccharomyces pombe

Abstract

Previously isolated Schizosaccharomyces pombe swi5 mutants are defective in mitotic mating-type switching and in repair of meiotic recombination-related DNA double-strand breaks. Here, we identify the swi5 gene, which encodes an 85-amino-acid polypeptide, similar to Sae3 of Saccharomyces cerevisiae, with an N-terminal predicted coiled-coil domain. A swi5 complete deletion mutant had normal mitotic growth rate but was hypersensitive to DNA-damaging agents and defective in mating-type switching. In meiosis, recombinant frequencies were reduced by a factor of ∼10. The swi5 deletion strongly reduced the viable spore yields of mutants lacking Rhp55 or Rhp57, proteins thought to aid joint molecule formation. Furthermore, the swi5 deletion strongly suppressed the low viable spore yield of mutants lacking Mus81•Eme1, which resolves joint molecules such as Holliday junctions. These and previous results indicate that the small Swi5 polypeptide acts in a branched pathway of joint molecule formation to repair meiotic DNA breaks.

THE repair of DNA strand breaks is crucial for the life of cells, since unrepaired broken DNA is likely to misegregate at cell division and lead to aneuploidy and consequent sickness or death. DNA breaks can arise from accidents during replication or transcription or from exogenous DNA-damaging agents such as UV or ionizing irradiation. The faithful repair of DNA double-strand (ds) breaks appears to occur most frequently by homologous recombination, which requires the formation of joint molecules between the broken DNA and an intact homolog. In special circumstances cells use programmed DNA breaks to stimulate recombination. For example, in the fission yeast Schizosaccharomyces pombe and the budding yeast Saccharomyces cerevisiae both mating-type switching and meiotic recombination are initiated by programmed DNA breaks (Haber 1998b; Davis and Smith 2001; Arcangioli and Thon 2003). As expected, certain mutants of these species are deficient in both processes and in the repair of damaged DNA.

Ten linkage groups of mutations that reduce the frequency of mating-type switching in S. pombe have been described (Gutz and Schmidt 1985). Mutants altered in one of these groups, swi5, are also hypersensitive to UV and ionizing irradiation and deficient in meiotic recombination (Schmidt et al. 1987, 1989). During meiosis swi5 mutants make DNA breaks but do not efficiently repair them (Young et al. 2004). During mitotic growth a DNA lesion at the mat1 locus that is essential for mating-type switching is made at normal level in swi5 mutants, but this lesion only inefficiently leads to switching (Egel et al. 1984). Thus, Swi5 is important for DNA break repair in both meiosis and mitosis.

Repair of DNA breaks by homologous recombination proceeds in steps via a joint molecule intermediate. Single-strand (ss) DNA is produced at the break via DNA unwinding by a helicase or digestion by a nuclease. The ss DNA pairs and undergoes strand exchange with an intact homologous DNA molecule to form a joint molecule, such as a D-loop or a Holliday junction. This step is promoted by bacterial RecA protein or its eukaryotic homolog Rad51; these proteins are aided by others, such as S. cerevisiae Rad52, Rad55, and Rad57 (Sung et al. 2000). Joint molecules are resolved into separate molecules by special enzymes; for example, Holliday junctions are resolved by the bacterial RuvC protein or, in S. pombe, the Mus81·Eme1 complex (Boddy et al. 2001; Gaillard et al. 2003; Osman et al. 2003). Remaining ss nicks and gaps are sealed by DNA polymerases and ligases.

We report here the identification of swi5, which encodes a remarkably small protein. Akamatsu et al. (2003) independently identified the swi5 gene and studied its role in mitotic DNA repair. Results presented here indicate that Swi5 is involved in DNA strand exchange, i.e., before joint molecule resolution by Mus81·Eme1 during meiotic recombination.

MATERIALS AND METHODS

Strains and plasmids:

Strains and their genotypes are listed in Table 1. Plasmids pCE1 and pCE2 were constructed by inserting into the EcoRI site of plasmid pFY20 (Li et al. 1997) an EcoRI-digested product of a polymerase chain reaction (PCR). The PCR used DNA from strain GP3652 as template and primers 32 or 35 nucleotides long, with 5′-terminal EcoRI sites, designed to amplify the 1.2-kb region from bp 2699–3856 of cosmid SPBC409 (GenBank accession no. AL109822). Methods were as described in Ausubel et al. (2003). Plasmids were introduced into S. pombe cells by LiOAc-stimulated transformation as described by Ito et al. (1983).

TABLE 1.

S. pombe strains

Strain	Genotype	Source or reference
GP13	h− ade6-52	Ponticelli and Smith (1989)
GP24	h+ ade6-M26	Ponticelli and Smith (1989)
GP50	h90	A. Ponticellia
GP289	h− ade6-52 rec10-109	Ponticelli and Smith (1989)
GP583	h− ade6-52 swi5-134	DeVeaux et al. (1992)
GP584	h+ ade6-M26 swi5-134	DeVeaux et al. (1992)
GP788	h+ ade6-M26 ura1-61	L. DeVeauxa
GP814	h− ade6-52 swi5-134 ura4-595	J. Virgina
GP1908	h− ade6-M26 ura4-D18 lys4-95	F. Lia
GP3652	h− smt0 ade6-M26 rad50::kanMX pat1-114	Young et al. (2004)
GP3999	h− ade6-52 swi5-201::kanMX6	Transformationb of GP13
GP4018	h+ ade6-52 swi5-201::kanMX6	GP24 × GP3999
GP4019	h− ade6-M26 swi5-201::kanMX6	GP24 × GP3999
GP4020	h+ ade6-M26 swi5-201::kanMX6	GP24 × GP3999
GP4075	h+ ade6-3049 lys3-37 pro1-1	This studya
GP4165	h− ade6-52 lys3-37 pro1-1	GP289 × GP4075
GP4196	h90 ade6-M26 swi5-134	GP50 × GP584
GP4197	h90 ade6-M26 swi5-201::kanMX6	GP50 × GP4020
GP4213	h+ ade6-M26 swi5-201::kanMX6 ura1-61	GP788 × GP4019
GP4216	h− ade6-52 swi5-201::kanMX6 lys3-37 pro1-1	GP4018 × GP4165
GP4367	h+ ade6-M26 swi5-134 ura4-D18	GP584 × GP1908
HE415	h+ lys1 ade4-31	Schmidt (1993)
HE416	h+ lys1 ade4-31 swi5-39	Schmidt (1993)
HE424	h− ura1-171 his6-365	Schmidt (1993)
HE426	h− ura1-171 his6-365 swi5-39	Schmidt (1993)
HE596	h− ade4-31 swi5::ura4+	This study
HE597	h+ lys3-131 swi5::ura4+	This study

×, meiotic cross.

^aGenealogy available upon request.

^bTransforming DNA was a PCR product using pFA6a-kanMX6 as template (Bahler et al. 1998). See materials and methods.

Construction of swi5-201::kanMX6:

The method of Bähler et al. (1998) was used. Primers 100 nucleotides long and plasmid pFA6A-kanMX6 as template were used in a PCR to generate a 1.6-kb DNA fragment with the kanMX6 cassette flanked by 80 bp identical to the DNA immediately adjacent to the swi5 gene (SPBC409.03; bp 3085–3164 and 3640–3719 on cosmid SPBC409). This fragment was purified by gel electrophoresis, extraction with a QIAquick kit (QIAGEN, Chatsworth, CA), and precipitation with ethanol. Approximately 1 μg was used to transform ∼4 × 10⁷ cells of strain GP13, with selection for G418 antibiotic resistance. Colonies were purified and tested for stability of resistance, and the mutation verified by nucleotide sequencing using primers ∼200 bp outside the substituted region.

Culture media and meiotic crosses:

Rich yeast extract media (liquid YEL and solid YEA), Edinburgh minimal media (liquid EMM2 and solid EMM2 agar), malt extract agar (MEA), and sporulation agar (SPA) are described by Gutz et al. (1974). Media were supplemented with adenine and other nutrients (100 μg/ml) as required. For meiotic crosses the two parental strains were grown to saturation in YEL and 100 μl (50 μl for Table 5) of each culture were mixed. The cells were collected by centrifugation, washed twice in 1 ml of H₂O, suspended in ∼15 μl of H₂O, and deposited on SPA (MEA for Table 2). After 2 days at 25° spores were harvested and assayed for total viable spores and recombinants as described by Ponticelli and Smith (1989), except that EtOH treatment was for 10–15 min and Ade⁺ recombinant frequencies were determined by plating on YEA with and without guanine (200 μg/ml). Crosses between strains GP814 and GP4367 (Table 3) were plated on EMM2 plus uracil with or without adenine; uracil was omitted for crosses involving a ura4⁺ plasmid (pCE1, pCE2, or pFY20). For the experiments in Table 5 two colonies of each strain (h⁺ or h⁻) freshly grown on YEA + adenine were picked to separate tubes of YEL + adenine (5 ml) and grown to saturation; the four combinations of h⁺ × h⁻ matings, with control strains, were done and analyzed concurrently.

TABLE 5.

swi5 interacts withmus81 and oppositely withrhp55 andrhp57

		swi5+		swi5Δ
Mutanta	nb	Ade+/106 viable sporesc	Viable spores/cellc	Ade+/106 viable sporesc	Viable spores/cellc
+	12	3700 ± 210	≡100	270 ± 80	44
mus81Δ	8	ND	<0.01	250 ± 52	6.8
mus81Δ dmc1Δ	4	1200 ± 470	0.25	400 ± 23	6.5
dmc1Δ	8	1600 ± 95	81	470 ± 45	58
rhp55Δ	4	2100 ± 280	33	ND	<0.1
rhp57Δ	4	1200 ± 130	11	ND	<0.001

ND, not determined, as too few viable spores were produced for accurate measurement.

^aAlleles were mus81::kanMX6 (Boddy et al. 2001), dmc1::ura4⁺ (Fukushima et al. 2000), rhp55::ura4⁺ (Khasanov et al. 1999), rhp57::ura4⁺ (Tsutsui et al. 2000), and swi5-201::kanMX6 (materials and methods) and were homozygous mutant or wild type as indicated. One parent was ade6-M26, and the other ade6-52. Strains with rhp55Δ or rhp57Δ were h⁺ or h⁻ smt-0 (Styrkarsdottir et al. 1993), and those with rhp55Δ, rhp57Δ or dmc1Δ were ura4-D18 (Grimm et al. 1988). Complete genotypes and genealogies are available upon request.

^bNumber of independent matings analyzed for swi5⁺ and swi5Δ. Matings were in sets of four for each genotype, done on three different days for n = 12 and on two different days for n = 8 (see materials and methods).

^cMating and meiosis were on SPA at 25°. Ade⁺ recombinant frequencies are the means ± SEM. Viable spore yields are expressed relative to wild-type matings, which produced 8.5 ± 0.96 (n = 12) viable spores per viable cell (of the less numerous parent) added to the mating mixture. This yield is greater than the theoretical four spores per limiting haploid cell because of slight residual growth of cells on the sporulation medium.

TABLE 2.

lys1-ade4 intergenic recombination inswi5 mutants (tetrad analysis)

h− parent		h+ parent
HE no.	swi5	HE no.	swi5	PD	NPD	T	cMa
424	+	415	+	50b	55b	214b	∼260
426	39	416	39	129b	9b	103b	35
426	39	416	39	37	4	50	51
597	Δc	596	Δ	188	24	197	46

PD, parental ditype; NPD, nonparental ditype; T, tetratype.

^aGenetic distance in centimorgans was calculated for the wild type from the physical distance (1.63 Mb) between lys1 and ade4 (Wood et al. 2002) and the genome mean of 0.16 cM/kb (Young et al. 2002) and for the swi5 mutants using the equation of Haldane (1919).

^bData from Schmidt (1993).

^cswi5::ura4⁺.

TABLE 3.

ade6intragenic recombination inswi5 mutants

ade6-52 parent			ade6-M26 parent
GP no.a	swi5	Plasmid	GP no.	swi5	Ade+ recombinants per 106 viable sporesb
13	+	—	24	+	3500 ± 500
3999	swi5 Δc	—	24	+	2700 ± 430
583	swi5-134	—	584	swi5-134	680 ± 80
3999	swi5 Δ	—	584	swi5-134	520 ± 80
3999	swi5 Δ	—	4020	swi5 Δ	280 ± 30
814	swi5-134	—	4367	swi5-134	50 ± 13d
814	swi5-134	pCE1e	4367	swi5-134	3400 ± 650f
814	swi5-134	pCE2e	4367	swi5-134	4000 ± 860f
814	swi5-134	pFY20	4367	swi5-134	530 ± 120f

—, no plasmid present.

^aFor plasmid-containing strains, the GP number indicates the parent strain without a plasmid.

^bThe recombinant frequencies in three independent matings are given as the mean ± SEM.

^cswi5-201::kanMX6.

^dThe lower frequency in these crosses than in crosses between GP583 and GP584 appears to stem, for unknown reasons, from the ura4 mutation in strains GP814 and GP4367 (data not shown).

^eContains a 1.2-kb swi5 fragment in opposite orientations in the vector pFY20 (see materials and methods).

^fThe recombinant frequencies of two experiments with two independent transformants are given (n = 4).

RESULTS

Cloning the swi5 gene and sequencing swi5 mutations:

Using the iodine-positive reaction of h⁹⁰ swi⁺ colonies (Gutz et al. 1974), we obtained from an S. pombe genomic library (Fleck et al. 1992) three independent clones that complemented the swi5-39 mutation, initially characterized by Gutz and Schmidt (1985)(data not shown). Restriction digestion and nucleotide sequencing revealed that these clones, pHE12, pHE13, and pHE16, have at least three putative genes in common; the largest of these genes is SPBC1709.19c (Figure 1). Substitution of the 1.8-kb HindIII fragment containing ura4⁺ (Grimm et al. 1988) for 2.8 kb of SPBC1709.19c and its flanking DNA created an allele, designated swi5::ura4⁺, that conferred UV sensitivity, mating-type switching deficiency, and meiotic recombination deficiency comparable to that of swi5 mutants (Table 2 and data not shown). Unlike the previously described swi5 mutants, however, strains with the swi5::ura4⁺ allele grew slowly; furthermore, a subclone, designated pHE12-S1.1 and containing all of SPBC1709.19c (Figure 1), failed to complement swi5-39 (data not shown). These results indicated that the swi5 gene was on the initial clones but was not SPBC1709.19c.

Figure 1.—

Structure of the swi5 gene and its surroundings on chromosome II of S. pombe and a comparison of Swi5 and S. cerevisiae Sae3. The top line shows a 9-kb segment of chromosome II contained on the overlapping cosmids SPBC1709 and SPBC409; boxes with arrows indicate genes and their direction of transcription. swi5 is identical to SPBC409.03 (see results). Lines below the top line indicate the extents of five clones and one deletion (swi5::ura4⁺) and their ability to complement the swi5-39 mutation (left column). The expanded line in the middle of the figure represents swi5 and the two flanking genes. Open boxes indicate open reading frames (exons). Below this line are the swi5-201::kanMX6 substitution and the positions of the swi5 point mutations. At the bottom of the figure is a ClustalW alignment, with high gap penalty, of the deduced S. pombe Swi5 and S. cerevisiae Sae3 proteins (see discussion). Vertical bars indicate 14 identical amino acids, and the horizontal bar over the Swi5 sequence indicates a predicted coiled-coil structure (Lupas 1996).

On the basis of the information above, we determined the nucleotide sequence of the region flanking SPBC1709.19c in the previously described swi5 mutants. The swi5-134 and swi5-147 mutations (Gutz and Schmidt 1985) were identical G → A changes 475 bp to the 5′ side of SPBC1709.19c, and swi5-39 was C → T 795 bp to the 5′ side (Figure 1). We noted that these mutations were in a hypothetical small gene with two introns. A cDNA corresponding to part of the conceptual mRNA from this gene has been reported (GenBank accession no. AU013615; M. Morimyo and K. Mito, unpublished data), supporting the notion that this hypothetical gene is active and that its RNA transcript is spliced as predicted. If so, the swi5-134 and 147 mutations would change the beginning of the 5′ splice site of the first intron from the invariant 5′ GT … to 5′ AT …, and swi5-39 would change codon 38 from CAA (Gln) to UAA (nonsense; Figure 1). The gene SPBC409.03, now designated swi5, is predicted to encode an 85-amino-acid polypeptide whose N-terminal ∼30 amino acids are predicted to form a coiled-coil (Lupas 1996). Akamatsu et al. (2003) independently identified swi5 as SPBC409.03.

Deletion of the swi5 coding sequence and the null phenotype:

On the basis of the above information, we deleted precisely the coding sequence of SPBC409.03 and replaced it with the kanMX6 cassette to create the swi5-201::kanMX6 allele, called here swi5Δ. Strains with this allele had a phenotype much like that of previously described swi5 alleles. Mitotic growth was not significantly different from that of wild type (data not shown). Mating-type switching was much reduced and produced an iodine-staining reaction, indicative of mating and sporulation (Gutz et al. 1974), comparable to that of swi5-134 (data not shown). The deletion mutant was slightly more sensitive to UV irradiation than was swi5-134, but both mutants were considerably more sensitive than wild type (data not shown), as reported by Akamatsu et al. (2003).

To quantitatively evaluate complementation between swi5 alleles, we measured meiotic recombination in ade6 intragenic crosses (Table 3). Gene conversion at ade6 was reduced slightly more by the swi5Δ substitution than by swi5-134 (factors of 12 and 5, respectively, relative to wild type). These two alleles failed to complement, but two subclones, pCE1 and pCE2, containing the swi5 coding sequence and all of the DNA between the two flanking genes (Figure 1), did complement swi5-134 (Table 3). These results indicate that SPBC409.03 is swi5.

Crossing over was also reduced by a swi5 deletion to about the same extent as by a swi5 point mutation. By tetrad analysis, crossing over between lys1 and ade4 was reduced from ∼260 cM in wild type to ∼35–50 cM in swi5 mutants, a reduction by a factor of ∼6 (Table 2). By random spore analysis, recombination in the lys3–ura1 and ura1–pro1 intervals was reduced in the swi5Δ mutant by factors of 37 and 8, respectively (Table 4). The swi5-39 mutation reduces crossing over in six additional intervals by factors of 4–17 (Schmidt 1993). In summary, swi5 deletion mutants have phenotypes similar to those of the point mutants.

TABLE 4.

Recombination inswi5 mutants (random spore analysis)

h− parenta		h+ parenta		% recombinantsb
GP no.	swi5	GP no.	swi5	lys3–ura1	ura1–pro1	Ade+ recombinants/106 viable spores
4165	+	788	+	14.7	27.6	4400
4216	Δc	4213	Δ	0.4	3.4	320

^aThe h⁻ parent was ade6-52 lys3-37 pro1-1; the h⁺ parent was ade6-M26 ura1-61.

^bFrom each cross 232 segregants were tested; reciprocal types were nearly equally frequent.

^cswi5-201::kanMX6.

A swi5 deletion suppresses the low viable spore yield of mus81 mutants:

The Mus81·Eme1 endonuclease appears to be a Holliday junction resolvase essential for meiotic crossing over (Boddy et al. 2001; Gaillard et al. 2003; Osman et al. 2003; Smith et al. 2003). mus81 mutants have very low viable spore yields, presumably because joint molecules are formed but remain unresolved in the absence of this endonuclease. The low viable spore yield is suppressed by a rec6 or rec12 mutation, which blocks meiotic DNA breakage and presumably joint molecule formation (Cervantes et al. 2000; Boddy et al. 2001; Young et al. 2002; Osman et al. 2003). Because swi5 mutants, like rec6 and rec12 mutants, are deficient in meiotic recombination but have relatively high viable spore yields (Schmidt et al. 1987; Ponticelli and Smith 1989; DeVeaux et al. 1992; Schmidt 1993; Lin and Smith 1994; Young et al. 2004), we tested the ability of a swi5Δ mutation to suppress the low viable spore yield of a mus81 mutation. The mus81Δ mutation decreased the viable spore yield, relative to that of wild type, by a factor of >10,000; this low level was increased >700-fold by swi5Δ (Table 5). Although this suppression was strong, it was not complete, as the viable spore yield was lower than that of swi5Δ by a factor of ∼6. These results suggest that Swi5 and perhaps other factors act before Mus81·Eme1. The Ade⁺ recombinant frequency was indistinguishable in swi5Δ and swi5Δ mus81Δ strains, indicating that Mus81·Eme1 has little effect on intragenic recombination (gene conversion) in swi5Δ mutants, as in swi5⁺ strains (Osman et al. 2003; Smith et al. 2003).

Viable spore yields are reduced synergistically by swi5 and an rhp55 or rhp57 mutation but not a dmc1 mutation:

The accumulation of broken DNA in a swi5 mutant (Young et al. 2004) and the suppression of mus81Δ by swi5Δ (Table 5) suggest that Swi5 acts during the DNA strand-exchange step of recombination. Other proteins have been inferred to act at this step; these proteins include the DNA strand-exchange factors Rad51 and Dmc1 and the accessory factors Rhp55 and Rhp57 (Grishchuk and Kohli 2003 and references therein). This inference suggests that there might be genetic interactions among mutations in the genes encoding these proteins. To test this possibility, we measured meiotic recombination and viable spore yields in single and double mutants.

We first tested interaction of swi5Δ and dmc1Δ. Intragenic ade6 recombination was reduced by a factor of ∼2 by dmc1Δ and by a factor of ∼14 by swi5Δ (Table 5). In the swi5Δ dmc1Δ double mutant, recombination was intermediate between the levels in the single mutants. Viable spore yields were only modestly reduced, by a factor of ∼2 or less in these mutants, but again the phenotype of the double swi5Δ dmc1Δ mutant was intermediate between those of the single mutants.

The similar phenotypes of swi5Δ, dmc1Δ, and swi5Δ dmc1Δ mutants suggest that Swi5 and Dmc1 act at closely related steps of meiosis. We therefore tested interaction of dmc1Δ with mus81Δ. Like swi5Δ, the dmc1Δ mutation increased the viable spore yield of mus81Δ, by a factor of >25 but below the level of mus81Δ swi5Δ. The ade6 recombinant frequency in the dmc1Δ mus81Δ mutant was near that of the dmc1Δ single mutant (Table 5). The triple mutant mus81Δ swi5Δ dmc1Δ had a phenotype similar to that of the mus81Δ swi5Δ double mutant. These results support the suggestion that Swi5 and Dmc1 act at closely related steps before Mus81·Eme1.

In contrast to the swi5Δ dmc1Δ double mutant, which had a phenotype similar to that of the single mutants, double mutants with swi5Δ and either rhp55Δ or rhp57Δ had a much stronger phenotype than the single mutants. Viable spore yields were reduced, relative to that of wild type, by factors of ∼2–10 in the three single mutants (Table 5). The addition of the swi5Δ mutation to the rhp55Δ or rhp57Δ mutant reduced the viable spore yields by factors of >300 and >10,000, respectively. These yields were too low to allow a reliable measure of recombination. These results indicate that Swi5 acts at a step different from that of Rhp55 and Rhp57.

In summary, the swi5Δ mutation interacted strongly with mus81Δ, rhp55Δ, and rhp57Δ mutations, but only weakly with dmc1Δ. dmc1Δ also interacted with mus81Δ, but less strongly than did swi5Δ. Below, we discuss interpretations of these results.

DISCUSSION

Swi5 is a remarkably small protein—85 amino acids—that appears, from the evidence discussed below, to act with other proteins to form joint DNA molecules. This activity is important in the repair of single- and double-strand DNA breaks during both mitotic growth and meiosis.

Swi5 is related to the small Sae3 protein of S. cerevisiae (Figure 1). sae3 mutants were identified among sporulation-deficient mutants that are rescued by a spo11 mutation, a phenotype consistent with Sae3 acting after DNA break formation by Spo11 (McKee and Kleckner 1997). SAE3 was thought to encode a 50-amino-acid polypeptide from a single exon (McKee and Kleckner 1997), but further analysis suggests an additional exon and a protein of 91 amino acids (A. Shinohara, personal communication). This longer version has 14 amino acids identical to those of Swi5 scattered throughout the proteins (Figure 1). Physical analyses show that Sae3 and Swi5 are required for the repair of meiotic DNA double-strand breaks (McKee and Kleckner 1997; Young et al. 2004) and appear to act in concert with RecA-like strand exchange proteins. S. cerevisiae sae3 mutants have a phenotype indistinguishable from that of dmc1 mutants: both are meiosis specific, accumulate broken DNA hyperresected at the ends, and form less recombinant DNA than wild type does; the sae3 dmc1 double mutant has the same phenotype as the single mutants (McKee and Kleckner 1997). These results suggest that Sae3 and Dmc1, a meiosis-specific DNA strand-exchange protein (Bishop et al. 1992; Sehorn et al. 2004), act in the same step in DNA repair, the formation of joint molecules.

There are differences in the roles of Swi5 and Sae3, however. Although Sae3 is meiosis specific, Swi5 is required during mitotic growth for repair of DNA damage (Schmidt et al. 1989; Akamatsu et al. 2003; data not shown) and for mating-type switching (Egel et al. 1984). In addition, although the phenotypes of S. cerevisiae sae3 and dmc1 mutants are indistinguishable, the phenotypes of S. pombe swi5 and dmc1 mutants differ: during meiosis broken DNA accumulates in swi5 mutants but not in dmc1 mutants (Young et al. 2004), spore viabilities are lower in swi5 mutants than in dmc1 mutants (Grishchuk and Kohli 2003; Young et al. 2004; Table 5), and intragenic recombinant frequencies are lower in swi5 mutants than in dmc1 mutants (Schmidt et al. 1987; DeVeaux et al. 1992; Schmidt 1993; Fukushima et al. 2000; Grishchuk and Kohli 2003; Table 5). Thus, S. pombe Swi5 and Dmc1 are not functionally identical, although S. cerevisiae Sae3 and Dmc1 appear to be. Below, we discuss the roles of these and related proteins in meiotic and mitotic DNA break repair.

Meiotic DNA break repair:

During meiosis in S. pombe ds DNA breaks are made at high frequency at widely separated sites in the genome (Cervantes et al. 2000; Young et al. 2002). Formation of these breaks requires Rec12, a homolog of S. cerevisiae Spo11, which contains the active site for breakage (Keeney et al. 1997); at least nine additional proteins are also needed for high-frequency breakage (Cervantes et al. 2000; Young et al. 2002, 2004). Repair of these breaks requires several identified proteins, including Rad32, Rad50, Swi5, and Rad51 (Young et al. 2004). Rad32·Rad50·Nbs1 is a homolog of the S. cerevisiae Mre11·Rad50·Xrs2 complex, which is required to remove Spo11 covalently linked to the 5′ end of the broken DNA and to form ss DNA tails (Haber 1998a). Rad51 is a homolog of the S. cerevisiae Rad51 protein, which, like Escherichia coli RecA, promotes homologous DNA strand exchange between ss and ds DNA to form joint molecules (Sung et al. 2000). Optimal activity of Rad51, however, requires additional proteins, such as Rad54, Rad55, and Rad57. Rad55 and Rad57 form a tight heterodimeric complex and share limited amino acid sequence identity with Rad51; the Rad55·Rad57 complex may aid Rad51 by forming a complex with Rad51 (Sung et al. 2000). In S. pombe, a rad51 null mutation has a stronger meiotic phenotype than those of rhp55, rhp57, or dmc1 with respect to spore viability and recombinant frequencies (Grishchuk and Kohli 2003). Double-mutant analyses suggest that Rhp55 and Rhp57 act on one branch of a DNA repair pathway and Dmc1 on another; both of these branches are proposed to require Rad51 (Grishchuk and Kohli 2003).

Our results (Table 5) support the proposal of a pathway of meiotic DNA break repair with two branches leading to joint molecules (Figure 2). From the similar phenotypes of swi5Δ, dmc1Δ, and swi5Δ dmc1Δ mutants we infer that Swi5 and Dmc1 act in one branch of the pathway. From the strong synergisms between swi5Δ and rhp55Δ and between swi5Δ and rhp57Δ, we infer that Rhp55 and Rhp57 act in another branch. The Mus81·Eme1 endonuclease is required to resolve joint molecules, presumably made by Rad51, Dmc1, Swi5, Rhp55, and Rhp57, into crossover recombinants; in mus81Δ mutants very few viable spores are produced (Boddy et al. 2001; Gaillard et al. 2003; Osman et al. 2003: Smith et al. 2003; Table 5). Since swi5Δ and dmc1Δ suppress the low viable spore yield of mus81Δ mutants, we infer that Swi5 and Dmc1 act before Mus81. These inferences are concordant with current knowledge of the enzymatic activities of the proteins and indicate that Swi5 and Dmc1, and perhaps Rhp55 and Rhp57 as well, lead to joint molecules that are resolved by Mus81·Eme1.

Figure 2.—

Proposed branched pathway for meiotic recombination in S. pombe. After meiotic DNA replication, Rec12, aided by other proteins, makes DNA breaks independently of Rad32·Rad50·Nbs1 (the MRN complex; Cervantes et al. 2000; Young et al. 2004). The MRN complex, perhaps in conjunction with other proteins, resects one DNA strand. The resultant ss DNA forms a joint molecule with an intact homolog by either of two branches—one promoted by Swi5 plus Dmc1 and the other by Rhp55 plus Rhp57—each of which may require Rad51. The Swi5-Dmc1-dependent joint molecules are resolved by Mus81·Eme1 into recombinant chromatids (Boddy et al. 2001; Osman et al. 2003; Smith et al. 2003). Functions required for joint molecule resolution by the Rhp55-Rhp57-dependent branch have not been determined. This pathway is similar to one proposed by Grishchuk and Kohli (2003) and relies in part upon observations of meiotic recombination in S. cerevisiae (Roeder 1997).

Mitotic DNA break repair and mating-type switching:

During mitotic growth, the recovery from DNA damage is aided by several proteins. Mutant analyses suggest that Rad51, Rhp55, Rhp57, and Swi5 have relationships similar to those in meiosis. For example, rhp57 and swi5 mutants are mildly sensitive to UV and gamma irradiation, but the double mutant is as sensitive as rad51 or rad51 swi5 (Akamatsu et al. 2003; data not shown). Although the nature of the UV- and gamma-ray-induced lesions is not certain, these lesions may be primarily ss and ds breaks, respectively; if so, Swi5, as well as Rhp57 and perhaps Rhp55, would seem to aid Rad51 in the repair of both types of breaks.

Mating-type switching in mitotic S. pombe cells requires Swi5 and is initiated by a ss lesion at the mat1 locus; this lesion may be converted into a ds break during replication (Arcangioli and Thon 2003; Kaykov et al. 2004). Single-strand DNA with a 3′ end at the mat1 lesion presumably undergoes strand exchange with homologous DNA flanking the mat2 or mat3 locus; DNA synthesis primed by this end produces a copy of mat2 or mat3, which replaces the copy at mat1 (Arcangioli and de Lahondes 2000). Thus, mating-type switching is a type of copy-choice recombination, which relies on joint molecule formation by DNA strand transfer, as do break-join and break-copy recombination. swi5 mutants have normal levels of the mat1 lesion but rarely switch mating type. The inability of swi5 mutants to use the lesion to promote switching is consistent with the Swi5 protein being required for high-level DNA strand exchange, i.e., joint molecule formation. The mechanism of resolution of these joint molecules is unclear, since mating-type switching is not notably affected by mus81Δ (our unpublished observations). Nevertheless, we infer that the primary function of Swi5 is the same in meiotic and mitotic cells—the formation of joint molecules during DNA break repair.