Genetic Control of Recombination Partner Preference in Yeast Meiosis: Isolation and Characterization of Mutants Elevated for Meiotic Unequal Sister-Chromatid Recombination

Corresponding author: Dawn A. Thompson, University of California, Department of Physiology, 513 Parnassus Ave., Rm. S-762, Box 0444, San Francisco, CA 94143-0444., dthmpson{at}cgl.ucsf.edu (E-mail)

Communicating editor: P. J. PUKKILA

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Meiotic exchange occurs preferentially between homologous chromatids, in contrast to mitotic recombination, which occurs primarily between sister chromatids. To identify functions that direct meiotic recombination events to homologues, we screened for mutants exhibiting an increase in meiotic unequal sister-chromatid recombination (SCR). The msc (meiotic sister-chromatid recombination) mutants were quantified in spo13 meiosis with respect to meiotic unequal SCR frequency, disome segregation pattern, sporulation frequency, and spore viability. Analysis of the msc mutants according to these criteria defines three classes. Mutants with a class I phenotype identified new alleles of the meiosis-specific genes RED1 and MEK1, the DNA damage checkpoint genes RAD24 and MEC3, and a previously unknown gene, MSC6. The genes RED1, MEK1, RAD24, RAD17, and MEC1 are required for meiotic prophase arrest induced by a dmc1 mutation, which defines a meiotic recombination checkpoint. Meiotic unequal SCR was also elevated in a rad17 mutant. Our observation that meiotic unequal SCR is elevated in meiotic recombination checkpoint mutants suggests that, in addition to their proposed monitoring function, these checkpoint genes function to direct meiotic recombination events to homologues. The mutants in class II, including a dmc1 mutant, confer a dominant meiotic lethal phenotype in diploid SPO13 meiosis in our strain background, and they identify alleles of UBR1, INP52, BUD3, PET122, ELA1, and MSC1–MSC3. These results suggest that DMC1 functions to bias the repair of meiosis-specific double-strand breaks to homologues. We hypothesize that the genes identified by the class II mutants function in or are regulators of the DMC1-promoted interhomologue recombination pathway. Class III mutants may be elevated for rates of both SCR and homologue exchange.

MEIOSIS reduces the chromosome complement from diploidy to haploidy by a single round of DNA replication followed by two rounds of chromosome segregation. At the first meiotic division (MI), homologous chromosomes, which consist of pairs of sister chromatids, disjoin to opposite poles (reductional division). The second meiotic division (MII) resembles mitosis in that sister chromatids separate and segregate (equational division). For homologues to properly disjoin at MI, they must pair, recombine, and synapse. In MI prophase, homologous chromosomes align and pair with one another along their length. Pairing is followed by formation of the synaptonemal complex (SC; ALBINI and JONES 1987 ). SC formation initiates with the assembly of axial elements along the pairs of sister chromatids. A less densely staining central element then forms between the two homologues. In the completed (tripartite) SC, the axial elements are called lateral elements, and structures called transverse filaments extend from the central element to the lateral elements. The chromatin of each pair of sister chromatids is organized into loops attached at the base to the lateral elements (VON WETTSTEIN et al. 1984 ; HEYTING 1996 ). Synapsis is defined as the intimate association of homologues in the context of mature SC. At full synapsis, the entire structure (paired homologues plus SC) is called a meiotic bivalent.

In Saccharomyces cerevisiae, recombination is induced 100- to 1000-fold in meiosis, and most or all is initiated, concomitantly with SC formation, by meiosis-specific double-strand breaks (DSBs; reviewed in LICHTEN and GOLDMAN 1995 ). In this article, "exchange" refers to reciprocal events, and "recombination" refers to the sum of reciprocal and nonreciprocal events. Exchange between homologues in the context of mature SC (ENGEBRECHT et al. 1990 ) is required to form the stable interconnections, cytologically observed as chiasmata (CARPENTER 1988 ), that are necessary to orient the meiotic bivalent with respect to the MI spindle apparatus (reviewed in BASCOM-SLACK et al. 1997 ). Mutations that disrupt interhomologue exchange result in spore inviability because of missegregation of homologous chromosomes in MI. In addition, sister chromatids are closely associated with each other and with proteins of the axial elements when homologues are fully synapsed before the MI division (MOENS and PEARLMAN 1988 ). It has been proposed that this sister-chromatid cohesion is also necessary for chiasma function (MAGUIRE 1990 , MAGUIRE 1995 ). Mutations that disrupt sister-chromatid cohesion result in precocious separation of sister chromatids before the separation of homologues in MI (MIYAZAKI and ORR-WEAVER 1992 ; MOLNAR et al. 1995 ).

Although phenotypic analysis of meiotic mutants clearly indicates that chromosome pairing, recombination, and synapsis are interdependent, the exact relationship among these processes remains to be delineated. In yeast, it appears that early steps in the recombination pathway are required for synapsis, which initiates at the sites of recombination events (reviewed in ROEDER 1997 ). For example, mutants that are defective for meiotic recombination do not form SC. However, although early steps in the meiotic recombination pathway promote synapsis, the formation of recombinant products at normal levels depends on proper synapsis (reviewed in ROEDER 1997 ).

Meiotic exchanges occur preferentially between homologous chromatids (reviewed in PETES and PUKKILA 1995 ; KLECKNER 1996 ; ROEDER 1997 ). However, KADYCK and HARTWELL 1992 showed that DNA damage induced in G2 of the mitotic cell cycle was repaired preferentially by interaction with the sister chromatid. These observations indicate that as a cell enters meiosis, there is a change in recombination partner preference from intersister to interhomologue. This implies the existence of a meiotic machinery that directs the repair of meiosis-specific DSBs to homologues and/or away from sisters. Mutations inactivating this machinery would increase intersister recombination in meiosis and reduce, but not eliminate, interhomologue exchange.

Several screens have identified genes in yeast required for wild-type levels of meiotic recombination between homologues (reviewed in PETES et al. 1991 ; ROEDER 1997 ). The mutations identified in these screens can be generally classified into two groups: those that eliminate recombination and those that retain a significant level. In the former class are mutations in SPO11, which encodes a protein homologous to type II topoisomerases and is the catalytic subunit of the complex responsible for meiosis-specific DSBs (BERGERAT et al. 1997 ; KEENEY et al. 1997 ). RAD50 and several others have phenotypes implying involvement at an "early" stage in the meiotic recombination process (MALONE and ESPOSITO 1981 ; MALONE et al. 1991 ; KLAPHOLZ et al. 1985 ). These mutants do not form meiosis-specific DSBs or SC, but they do proceed through the two divisions of meiosis (ALANI et al. 1990 ; CAO et al. 1990 ). In the absence of recombination, the homologous chromosomes missegregate at MI, resulting in aneuploid meiotic products that are largely inviable.

There are several mutations that reduce meiotic interhomologue recombination to 10–25% of the wild-type level. Possible candidates for genes encoding components of the machinery that biases the repair of meiosis-specific DSBs to homologous chromatids may be found in this group. Two of these, HOP1 and RED1, are meiosis-specific genes encoding axial/lateral element components (HOLLINGSWORTH and BYERS 1989 ; HOLLINGSWORTH et al. 1990 ; ROCKMILL and ROEDER 1990 ; SMITH and ROEDER 1997 ). MEK1/MRE4 encodes a putative meiosis-specific kinase (ROCKMILL and ROEDER 1991 ; LEEM and OGAWA 1992 ). Genetic evidence indicates that the products of these three genes interact to promote proper SC assembly (ROCKMILL and ROEDER 1990 , ROCKMILL and ROEDER 1991 ; HOLLINGSWORTH and JOHNSON 1993 ; HOLLINGSWORTH and PONTE 1997 ; FRIEDMAN et al. 1994 ), and this conclusion was supported by recent cytological studies (SMITH and ROEDER 1997 ; BAILIS and ROEDER 1998 ). In addition, the RED1/MEK1/HOP1 epistasis group is implicated in meiotic sister-chromatid cohesion. red1 mutants fail to form axial elements (ROCKMILL and ROEDER 1990 ) and are defective in meiotic sister-chromatid cohesion (BAILIS and ROEDER 1998 ). The defect in meiotic sister-chromatid cohesion may explain why the crossovers that do occur in this mutant are not effective in disjunction (ROCKMILL and ROEDER 1990 ). Phosphorylation of Red1p by Mek1p is required for meiotic sister-chromatid cohesion. hop1 mutants assemble axial elements, but synapsis is blocked (HOLLINGSWORTH and BYERS 1989 ; LOIDL et al. 1994 ). Although not absolutely required for axial element formation and sister-chromatid cohesion, Hop1p is required for proper Mek1p localization, and it appears to stabilize the Red1p and Mek1p interaction (BAILIS and ROEDER 1998 ). In addition, the interaction of Hop1p with Red1p is enhanced by the presence of MEK1 (DE LOS SANTOS and HOLLINGSWORTH 1999). Thus, all three genes are likely required to form functional axial elements capable of nucleating synapsis.

RAD51 and DMC1 encode ubiquitous and meiosis-specific recA homologues, respectively. In rad51 and dmc1 mutants, meiosis-specific DSBs occur at wild-type levels, but they are unrepaired and hyperresected, indicating that RAD51 and DMC1 are required for strand exchange during meiotic recombination (BISHOP et al. 1992 ; SHINOHARA et al. 1992 ). Chromosome pairing is delayed and incomplete in the two mutants (ROCKMILL et al. 1995 ). In addition, both mutants are delayed in synapsis, are reduced for meiotic recombination to 10% of the wild-type level, and can cause arrest in meiotic prophase subsequent to synapsis (BISHOP et al. 1992 ; ROCKMILL et al. 1995 ).

It has been proposed that one function of the SC-associated proteins encoded by HOP1, RED1, MEK1, and DMC1 is to bias meiotic recombination events to homologues (PETES and PUKKILA 1995 ; KLECKNER 1996 ; ROEDER 1997 ). A dmc1 mutant exhibits an increase in intrachromosomal recombination between directly repeated sequences (BISHOP et al. 1992 ). In addition, there is evidence that DMC1 functions in a meiotic recombination pathway that is biased toward interhomologue exchange and that this pathway has functions that are independent of those of the ubiquitous RAD51 pathway (DRESSER et al. 1997 ; SCHWACHA and KLECKNER 1997 ; SHINOHARA et al. 1997 ; ZENVIRTH et al. 1997 ). In a hop1 mutant, meiosis-specific DSBs are reduced to 10% of the wild-type level. Moreover, these DSBs are processed exclusively into intersister recombination intermediates (SCHWACHA and KLECKNER 1994 ). It has been postulated that meiotic sister-chromatid cohesion reduces the participation of sister chromatids in meiotic recombination events (SMITH and ROEDER 1997 ). This suggests that disruption of sister-chromatid cohesion in red1 and mek1 mutants would result in an increase in meiotic sister-chromatid recombination (see RESULTS and DISCUSSION).

RED1 and MEK1 are also required for the meiotic prophase arrest induced by a dmc1 mutation (XU et al. 1997 ), suggesting a link between meiotic sister-chromatid cohesion, recombination, and a surveillance system that monitors the faithful completion of meiotic recombination. The DNA damage checkpoint control genes RAD24, RAD17, and MEC1 (WEINERT et al. 1994 ) are also required for dmc1-induced arrest, which defines the meiotic recombination checkpoint (LYDALL et al. 1996 ). Spore viability is reduced in rad24, rad17, mec1-1, and mec3 mutants in a pattern indicative of a defect in homologue disjunction at MI (LYDALL and WEINERT 1995 ; LYDALL et al. 1996 ). This suggests that, in addition to the proposed monitoring function, these checkpoint genes have a role in ensuring the fidelity of interhomologue recombination and/or disjunction.

Although many individual functions required for the fidelity of meiotic recombination have been identified, a role in distinguishing sequences on homologues from those on sister chromatids, or other "ectopic" homology, has not yet been confirmed. This distinction is defined as partner choice, which results in an overall preference for homologues in meiotic recombination. We sought to identify components of the machinery that mediates proper meiotic recombination partner choice, using a screen designed specifically to detect mutants exhibiting an increase in meiotic unequal sister-chromatid recombination (SCR). We reasoned that, in recombination-competent mutants, loss of the preference for the homologue in meiotic recombination would be manifest as an increase in the frequency of SCR.

This approach has identified 38 mutants exhibiting the meiotic sister chromatid recombination-elevated phenotype (msc). The msc mutants were quantified with respect to meiotic unequal SCR frequency, disome segregation pattern, sporulation frequency, and spore viability in the one-division meiosis conferred by the spo13 allele. In addition, outcrossing the mutants to a SIR3 SPO13 congenic strain revealed a class that conferred a dominant meiotic lethal phenotype peculiar to our strain background. Analysis of the msc mutants according to these criteria defined three classes: Mutants with a class I phenotype identify new alleles of the meiosis-specific genes RED1 and MEK1, DNA damage checkpoint genes RAD24 and MEC3 (WEINERT et al. 1994 ), and a previously unidentified gene, MSC6. The dominant meiotic lethal class II mutants, which include a dmc1 mutant, identify alleles of UBR1, INP52, BUD3, PET122, ELA1, and MSC1-MSC3. Class III mutants, which identify alleles of MNR2 and MSC7, have characteristics consistent with a meiotic hyper-rec phenotype.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plasmid construction:
Plasmids were constructed using standard procedures (MANIATIS et al. 1982 ). The arg4::URA3 fusion gene in pDT113 was constructed as follows: YIP5 (STRUHL et al. 1979 ) was digested with PstI and AvaI, and the ends of the resulting 901-bp fragment containing the URA3 coding region were filled in with T4 DNA polymerase. In parallel, pMLC28::ARG4 (LEVINSON et al. 1984 ) was cut with SacI, treated with T4 DNA polymerase (New England Biolabs, Beverly, MA) to fill in the ends, and then cut with SnaBI to remove the 1558-bp fragment containing the ARG4 coding region. The resulting 4.4-kb vector fragment was ligated with the aforementioned 901-bp URA3 fragment to generate pDT113.

pMS12 was constructed by ligating the 3.5-kb SnaBI fragment containing a segment of chromosome VIII adjacent to the 3'-end of the ARG4 gene from pSPO13-1 (WANG et al. 1987 ) into the SnaBI site of pMLC28::ARG4.

pMS23 was constructed in several steps:

1. An ~1.4-kb, CUP1-containing BamHI fragment of pYep36::CUP1 (BUTT et al. 1984 ) was inserted into the BamHI site of pTZ18U (United States Biochemical, Cleveland) to generate pMS4.

2. pAB34 was cut with Sau3a, and the ends of the resulting 374-bp fragment containing ARSH4 were filled in with T4 DNA polymerase. This fragment was then inserted into the SmaI site of pMS4 to generate pMS5.

3. pDT113 was cut with PstI, and the ends were filled in with T4 DNA polymerase and then cut with EcoRV to liberate an ~1-kb fragment containing a 3'-segment of the arg4::URA3 fusion. In parallel, pMS5 was cut with PstI, and the ends were filled in with T4 DNA polymerase. The resulting pMS5 vector fragment was ligated to the aforementioned ~1-kb fragment containing a 3'-segment of arg4::URA3 to yield pMS6.

4. pDT113 was cut with PstI and BsaI, and the ends of the resulting ~2-kb fragment containing a 5'-segment of the arg4::URA3 fusion gene were made blunt with T4 DNA polymerase. In parallel, pMS6 was cut with SacI, and the 5'-overhang was removed with T4 DNA polymerase. The linear, blunt-ended product was then ligated to the aforementioned ~2-kb fragment containing a 5'-segment of the arg4::URA3 fusion gene to generate pMS7. The 5' and 3' arg4::URA3 fragments and the intervening CUP1 gene comprise the SCR construct.

5. pMS7 was digested with EcoRI and SphI, and the ends of the resulting ~5-kb fragment containing the SCR construct were made blunt with T4 DNA polymerase. In parallel, pMLC28::ARG4 was digested with HpaI to remove an ~2-kb fragment of the ARG4 gene. The resulting 4-kb fragment of pMLC28::ARG4 was ligated to the aforementioned ~5-kb fragment containing the SCR construct to generate pMS13.

6. pMS12 was digested with HpaI to liberate an ~3-kb fragment containing a segment of chromosome VIII adjacent to the 3'-end of ARG4, which was then ligated into the SnaBI site of pMS13 to generate pMS14.

7. A NotI linker was then inserted into the StuI site in the 5'-segment of the arg4::URA3 gene in pMS14 to generate pMS21.

8. A PmeI linker was then inserted into the XmnI site in the chromosome VIII ARG4 3'-segment in pMS21 to generate pMS22.

9. Finally, pASZ10 (STOTZ and LINDER 1990 ) was digested with BglII to liberate an ~2.5-kb ADE2-containing fragment. In parallel, pMS22 was digested with BglII to remove an ~1.5-kb fragment of the chromosome VIII ARG4 3'-segment. The resulting 11.7-kb pMS22 fragment was ligated to the aforementioned ~2.5-kb ADE2 fragment to generate pMS23.

pMS36 was constructed in several steps.

1. An ~1.4-kb, CUP1-containing BamHI fragment of pYep36::CUP1 was inserted into the BamHI site of pTZ18U to yield pMS4.

2. pAB34 was cut with Sau3a, and the ends of the resulting 374-bp fragment containing ARSH4 were filled in with T4 DNA polymerase. This fragment was then inserted into the SmaI site of pMS4 to yield pMS5.

3. pDT113 was digested with BsaI; the ends were filled in with T4 DNA polymerase and subsequently digested with EcoRV to liberate a 432-bp fragment containing the middle segment of the arg4::URA3 fusion gene. pMS5 was digested with PstI; the 5'-overhang was removed with T4 DNA polymerase and then ligated to the aforementioned 432-bp fragment containing the middle segment of the arg4::URA3 fusion gene to generate pMS8.

4. pDT113 was digested with PstI; the 5'-overhang was removed with T4 DNA polymerase, and the resulting fragment was then digested with EcoRV to liberate an ~1-kb fragment containing a 3'-segment of the arg4::URA3 fusion gene. In parallel, pMS8 was digested with SacI; the 5'-overhang was removed with T4 DNA polymerase and then ligated to the aforementioned ~1-kb fragment containing a 3'-segment of the arg4::URA3 fusion gene to generate pMS9.

5. pMS9 was then digested with KpnI and BamHI to remove the ARSH4 and CUP1 sequences. The resulting 4.3-kb fragment of pMS9 was treated with T4 DNA polymerase to make the ends blunt and then ligated to generate pMS10.

6. pMS10 was digested with BsaI; the 5'-overhang was removed with T4 DNA polymerase, and the resulting fragment was then digested with MscI to remove 373 bp of the 3'-segment of the arg4::URA3 gene. The resulting 4-kb fragment of pMS10 was ligated to generate pMS11. The tandem segments of the arg4::URA3 gene comprise the homologue homology (HH) construct.

7. pMS10 was digested with EcoRI and SphI, and the ends of the resulting ~1.1-kb fragment containing the HH construct were made blunt with T4 DNA polymerase. In parallel, pMLC28::ARG4 was digested with HpaI to remove an ~2-kb fragment of the ARG4 gene. The resulting 4-kb fragment of pMLC28::ARG4 was ligated to the aforementioned ~1.1-kb fragment containing the HH construct to generate pMS11.

8. pMS12 was digested with HpaI to liberate an ~3-kb fragment containing a segment of chromosome VIII adjacent to the 3'-end of ARG4, which was then ligated into the SnaBI site of pMS11 to generate pMS17.

9. A PmeI linker was then inserted into one of the two BglII sites in the chromosome VIII ARG4 3'-segment in pMS17 to generate pMS35.

10. Finally, pASZ10 was digested with BglII to liberate an ~2.5-kb fragment containing the ADE2 gene, which was ligated to BglII-digested pMS35 to generate pMS36.

pCP3 (FOSS and STAHL 1995 ) was digested with EcoRI and HindIII, and the ends of the resulting ~2.7-kb LYS2-containing fragment were filled in with T4 DNA polymerase. In parallel, pLG54 (GILBERTSON and STAHL 1996 ) was digested with BstEII and BglII to remove a 1.1-kb URA3-containing fragment. The ends of the resulting 4-kb pLG54 fragment were filled in with T4 DNA polymerase and then ligated to the aforementioned ~2.7-kb LYS2-containing fragment to generate pMS38.

pEF83 (FOSS and STAHL 1995 ) was digested with EcoRI; the 3'-overhangs were filled with with T4 DNA polymerase and then ligated to the 2-kb, ARG4-containing HpaI fragment of pMLC28::ARG4 to generate pMS39.

A 4.5-kb, SIR3-containing SalI fragment of pJR273 (obtained from George Sprague, Jr.) was ligated into XhoI/SalI-digested pRS306 (SIKORSKI and HIETER 1989 ) to yield pMS40. pMS41 was constructed by digesting pMS40 with NruI and ClaI to remove a 1.6-kb fragment of the SIR3 gene, making the ends blunt and inserting a PmeI linker.

pMS42 was constructed by ligating an ~1.1-kb, URA3-containing SmaI fragment from pJJ242 (JONES and PRAKASH 1990 ) into SmaI-digested pB84 (ROCKMILL and ROEDER 1990 ).

pMS43 was constructed by ligating the ~1.4-kb, KanMX4-containing BglII/EcoRV fragment of KanMX4 (WACH et al. 1994 ) into BglII/MscI-digested pRSQ303 (constructed by Joe Horeka).

pMS47 was constructed in several steps, beginning with filling in the 3'-overhangs of BglII-digested pMS43 with T4 DNA polymerase, and then the blunt ends were ligated to destroy the BglII site. The plasmid was then digested with ApaI, the 5'-overhangs were removed, and a BglII linker was inserted. The plasmid was then digested with NdeI, the 3'-overhangs were filled in with T4 DNA polymerase, and then the blunt ends were ligated to destroy the NdeI site. The plasmid was then digested with SalI, the 3'-overhangs were filled in with T4 DNA polymerase, and a NdeI linker was inserted. Insertion of the NdeI linker restores the SalI site.

pMS49 was made in several steps, beginning with ligating the ~1.1-kb, URA3-containing SmaI fragment from pJJ242 into NaeI-digested pMS47. The plasmid was then digested with EcoRI, and the ends were filled in with T4 DNA polymerase and then ligated to destroy the EcoRI site. The ~3.3-kb SPO13-containing BamHI/EcoRV fragment of YIP5::SPO13 (constructed by Larry Gilbertson) was then ligated to the BglII/EcoRV-digested plasmid.

Yeast strains:
Yeast strains were constructed and manipulated by standard genetic methods (SHERMAN et al. 1982 ). Yeast strains were transformed using a standard LiOAC procedure (ITO et al. 1983 ). The genotypes of the yeast strains used in this study are listed in Table 1. DT71was constructed in several steps: (1) DT 60.3a was transformed with BamHI/XhoI-digested pEF84 to introduce the GPA1-3'::TRP1 construct by one-step transplacement (ROTHSTEIN 1983 ), generating DT61; (2) DT60.3a was also tranformed with BamHI/XhoI-digested pEF154 to introduce the GPA1-3'::LEU2 construct by one-step transplacement, generating DT62; (3) DT61 was transformed with EcoRI/PmeI-digested pMS23 to introduce the (ADE2::SCR) construct by one-step transplacement, generating DT63; (4) DT62 was transformed with EcoRI/PmeI-digested pMS36 to introduce the (ADE2::HH) construct by one-step transplacement, generating DT64; (5) DT65 is a ura3::HIS3 segregant of a DT63 x DT47.1d cross; (6) DT65 was transformed with EcoRI/HindIII-digested pMS38 to introduce the spo13::LYS2 allele by one-step transplacement, generating DT66; (7) DT64 was transformed with EcoRI/HindIII-digested pMS38 to introduce the spo13::LYS2 allele by one-step transplacement, generating DT67; (8) DT68 (Table 1) is a segregant of a DT66 x Z140-51c cross (MALONEY and FOGEL 1980 ); (9) DT69 is a segregant of a DT67 x DT68 cross; (10) DT69 was transformed with BamHI/XhoI-digested pMS39 to transplace the GPA1-3'::LEU2 construct with a GPA1-3'::ARG4 derivative, generating DT70; (11) DT70 was transformed with XhoI-digested pMS41 to introduce the sir3 allele by two-step transplacement to yield DT71.

The following mutations were introduced into DT71. The red1::ADE2 allele was introduced by two-step transplacement with XhoI-digested pMS42. The hop1::LEU2 allele was introduced by one-step transplacement with BglII-digested pNH37-2 (HOLLINGSWORTH and BYERS 1989 ). The dmc1::LEU2 allele was introduced by one-step transplacement with XbaI-digested pNKY422 (BISHOP et al. 1992 ). The rad17::LEU2 allele was introduced by one-step transplacement with BamHI/XbaI-digested pWL8 (LYDALL and WEINERT 1995 ). The rad24::LEU2 allele was introduced by one-step transplacement with SmaI-digested pWL62 (LYDALL et al. 1996 ). The spo11::hisg allele was introduced by two-step transplacement with BglII/XbaI-digested pGB518 (C. N. GIROUX, unpublished results). The ubr1 allele was introduced by one-step transplacement with HindIII-digested pSOB30 (a gift from Alex Varshavsky). The inp52::LEU2 allele was introduced by one-step transplacement with a PCR product generated as described in STOLZ et al. 1997 , using pRS305 (SIKORSKI and HIETER 1989 ) as the template. The msc1::LEU2 allele was introduced by one-step transplacement with ApaI/BsaI-digested pMS82. The ydr205w::LEU2 allele was introduced by one-step transplacement with AscI-digested pMS84. The msc3::LEU2 allele was introduced by one-step transplacement with EcoRI-digested pMS85.

DT72 was obtained by sporulating DT71 and screening the spore colonies from dissected dyads for an aberrant segregant that was monosomic for the SCR-construct-containing derivative of chromosomeVIII (Fig 1). DT78 was constructed in several steps, beginning with the introduction of the SPO13 allele by two-step transplacement with EcoRI-digested pMS49. The SIR3 allele was then introduced by two-step transplacement with XhoI-digested pMS40. Mating type was then switched by transformation with pGAL-HO (HERSKOWITZ and JENSEN 1991 ), and transformants were tested for mating type. Southern blot analysis was used to verify the function of the SCR construct and the structure of all strains made by transformation.

View larger version (28K): In this window In a new window Download PPT slide   Figure 1. Experimental design. (A) Marker configuration on each homologue of the chromosome VIII disome in the sir3 spo13::LYS2 haploid strain, DT71, used to screen for mutants that do not prefer the homologue over the sister chromatid in meiotic recombination. The centromere-linked markers TRP1 and ARG4 allow the determination of the chromosome VIII meiotic segregation pattern. The Trp+ chromosome carries the SCR construct transplaced at the normal ARG4 locus. The SCR construct consists of a tandem pair of arg4::ura3 gene fragments with 432 bp of overlapping homology (shaded regions) separated by the CUP1 gene. The arg4::ura3-3', marked with an arrowhead, is missing sequences 3' of the shaded homologous region, while the arg4::ura3-5', marked with feathers, lacks the 5'-segment. The Arg+ chromosome carries the HH construct, which consists of a tandem pair of 432-bp segments, labeled arg4::ura3-m, that are homologous to the shaded regions in the SCR construct. (B) The three types of segregation occurring in a spo13 disomic haploid. (C) Two types of unequal SCR events (exchange and nonreciprocal gap repair) in the homologous segments of the arg4::ura3 segments (shaded areas) will generate an intact arg4::URA3 gene conferring a Ura+ phenotype. Both unequal SCR events also duplicate the CUP1 gene. The copper-resistance phenotype conferred by CUP1 is sensitive to copy number. The products of unequal SCR are selected on medium lacking uracil, tryptophan, and arginine + 240 µM CuSO4 · 5H2O. Interactions with the homologue cannot produce an intact arg4::URA3 gene.

The spo13 homozygous diploid strains were constructed by transforming the haploid disomic strains with the SIR3-containing plasmid pJR273 and crossing them to DT78. These diploids were sporulated, and haploid segregants of the appropriate genotype were mated.

Mutant screen protocol:
Mutagenesis was with the Tn3 transposon-mutagenized yeast genomic library constructed by BURNS et al. 1994 . DT71 was transformed with NotI-cleaved DNA from 15 pools of the yeast genomic library carrying random TN3::lacz::LEU2 insertions. A total of 53,523 individual Leu⁺ transformants were picked and patched onto YEPEG plates and grown for 3 days. The patches were then replicated to S-raffinose + 5-fluoroorotic acid (5-FOA, 0.1%) plates and grown for 2 days to select against mitotic unequal SCR recombinants. The patches were then replicated to SPO plates and incubated for 3 days. The centromere of the SCR-construct-containing chromosome VIII has been marked with TRP1 integrated 3' of the GPA1 locus located ~2 cM from CEN 8 and the centromere of the chromosome containing the HH construct with ARG4 at the equivalent location (Fig 1). To eliminate the contribution of mitotic loss events from the analysis, cells were selected that contained both a Trp⁺ chromosome that had experienced an unequal sister-chromatid recombination event, depicted in Fig 1C, and an Arg⁺ HH-containing chromosome. For example, mitotic loss of the Arg⁺ HH-containing chromosome would result in a frequency of meiotic unequal SCR comparable to that observed in the homologue strain (Table 2) and, thus, score positive in the screen. Cells of the desired genotype were selected by replica plating to a medium lacking tryptophan, arginine, and uracil and containing the appropriate concentration of copper sulfate (240 µM CuSO₄ · 5H₂O). These SD-Ura-Arg-Trp + 240 µM CuSO₄ · 5H₂O plates were incubated for 2 days, after which colonies were clearly visible. All incubations were at 30°. A total of 4 individual colonies from each of the 455 candidates displaying an increase in meiotic unequal SCR in the initial screen were rescreened for this phenotype, revealing 67 candidates in which at least 3 out of the 4 colonies exhibited an increase in meiotic unequal SCR comparable to that of a red1 mutant. For each putative mutant, dyads were dissected to determine the pattern of segregation of the chromosome VIII pair. Cells were incubated in 12% glusulase for 8 min at 25°, followed by 30 min on ice. The frequencies of reductional, equational, and aberrant segregations in each strain were determined by replica plating the spore colonies from the dissected dyads to medium lacking tryptophan and medium lacking arginine. Genomic DNA flanking the transposon insertion was recovered from each of the 38 candidates displaying a chromosome VIII segregation pattern differing significantly from that of DT71.

Plasmid rescue and DNA analysis:
Genomic DNA flanking the TN3::lacz::LEU2 insertion was cloned as described (BURNS et al. 1994 ) with the following modifications. Yeast strains were transformed with BamHI/NotI-digested pMS43 or pMS47, and transformants were selected on YEPD + G418 (200 µg/ml) plates (WACH et al. 1994 ). Integration into the TN3::lacz::LEU2 sequences replaces LEU2 with an ~1-kb EcoRI/HpaI fragment of the LEU2 gene. G418^r transformants were screened for correct integration of the rescue plasmid on medium lacking leucine. Genomic DNA from Leu^- transformants was isolated according to the Rapid DNA Isolation Protocol (HOFFMAN 1997 ) with the addition of one phenol and one chloroform-isoamyl alcohol extraction. Genomic DNA from pMS43 transformants was digested individually with EcoRI, XhoI, and SalI. Genomic DNA from pMS47 transformants was digested individually with EcoRI, XhoI, SalI, BglII, and NdeI and then ligated. The KanMX4 module confers resistance to 50 µg/ml kanamycin in Escherichia coli cells (WACH et al. 1994 ). The ligated DNA was used to transform E. coli strain XLII-Blue (Stratagene, La Jolla, CA), and kanamycin-resistant transformants were screened for plasmids bearing a chromosomal insert. A primer complementary to the lacZ fragment (NEB sequencing primer catalog no. 1224) was used to sequence the adjacent chromosomal insert. Sequencing was carried out at the Institute of Molecular Biology sequencing facility at the University of Oregon. The locus of transposon insertion was determined by reference to the Stanford S. cerevisiae Genome Database (http://genome-www.stanford.edu/Saccharomyces/).

Recombination assays:
Yeast strains were grown to saturation in 2-ml cultures of YEPEG. The entire culture was then used to inoculate 100 ml S-raffinose + 5-FOA (0.1%) and was incubated for ~36 hr to select against mitotic SCR recombinants. Cells were pelleted, washed twice with sterile water, and diluted 1:4 in liquid sporulation medium. Aliquots from the liquid sporulation cultures were washed twice in 250 mM EDTA, pH 8.0, followed by two washes with sterile H₂O, and then plated on SD-Ura-Arg-Trp + 240 µm CuSO₄ · 5H₂O and on YEPD medium to determine the mitotic unequal SCR frequency per viable cell. Cultures were aerated for 3 days to induce sporulation, and the meiotic unequal SCR frequency was determined as described above. All incubations were at 30°. At least three independant colonies were assayed for each strain.

Sporulation frequency and spore viability:
Sporulation frequency in liquid sporulation cultures was determined microscopically. Spore viability was determined by dissection of dyads from SPO plates that had been incubated for 3–4 days at 30°. At least 100 individual spores were analyzed for each strain.

Linkage analysis of the msc mutants:
Each of the msc mutants was transformed with the SIR3-containing plasmid pJR273 and subsequently crossed to DT78. For each cross, the spore colonies from at least 20 four-spore-viable tetrads were analyzed for growth on SD-Arg, SD-Trp, SD-Leu, and SD-Lys media. In all crosses producing live spores, the LEU2 marker segregated in a 2:2 pattern, indicating that these msc mutants were carrying a single-transposon insertion. Linkage of the msc phenotype to the transposon insertion was determined by assaying meiotic unequal SCR in at least four Leu⁺ and four Leu^- segregants of the appropriate genotype from each cross. In the class II mutants, which did not produce viable spores when crossed to DT78, linkage was tentatively assessed by deleting the ORF identified by the transposon insertion in DT71 and assaying meiotic unequal SCR in the resulting mutant.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of mutants defective in directing meiotic recombination events to homologous chromatids:
Yeast mutants defective in directing the repair of meiosis-specific DSBs to homologous chromatids were isolated using a screen based on the strategy developed by HOLLINGSWORTH and BYERS 1989 . They isolated mutants unaffected for intersister and/or intrachromatid recombination, but reduced for recombination between homologues.

We reasoned that, in meiotic recombination-competent mutants, loss of the preference for the homologue would be manifest as an increase in the frequency of meiotic SCR. To specifically detect an increase in meiotic SCR, we designed an SCR construct on the basis of those described in FASULLO and DAVIS 1987 . The mutations we were seeking were expected to result in an elevation of meiotic SCR at the expense of interhomologue exchange. Since mutations that reduce interhomologue exchange alter chromosome disjunction (reviewed in HAWLEY 1988 ), our putative mutants exhibiting an increase in meiotic unequal SCR were also screened for an alteration in chromosome disjunction (Fig 1 and Fig 2).

View larger version (26K): In this window In a new window Download PPT slide   Figure 2. Strategy and summary of the msc mutant screen.

The strain used in the screen, DT71, is a spo13 sir3 haploid, disomic for chromosome VIII. The haploidy facilitates isolation of recessive mutations. The sir3 mutation results in the derepression of the normally silent-mating-type loci HML and HMR, which leads to coexpression of a and (SHORE et al. 1984 ), resulting in a haploid strain competent to undergo meiosis. The spore inviability of mutants affecting interhomologue exchange can be rescued by a mutation in the SPO13 gene. Meiotic recombination occurs at wild-type levels in spo13 mutants, which then skip one meiotic division and produce dyads containing two viable spores (KLAPHOLZ et al. 1985 ). The elimination of one meiotic division serves to bypass the requirement for recombination and/or synapsis to produce viable spores in spo13 meiosis (KLAPHOLZ et al. 1985 ; ROCKMILL and ROEDER 1988 ; HOLLINGSWORTH and BYERS 1989 ; MALONE et al. 1991 ). Thus, spo13 mutations have been exploited in the characterization of mutations that affect these processes. In addition, the single-division meiosis in the spo13 mutant permits a haploid to sporulate and produce two viable spores (WAGSTAFF et al. 1982 ).

In spo13 disomic haploids, a homologous chromosome pair exhibits three types of segregation in the single-division meiosis: reductional (as in MI), equational (as in MII), and aberrant (one spore monosomic and one spore trisomic; WAGSTAFF et al. 1982 ; Fig 1B). In spo13 meiosis, the distribution of the three types of segregation appears to depend on the frequency of interhomologue exchange, chromosome pairing, and/or synapsis. Mutations that disrupt any or all of these processes result in a shift in favor of equational segregation (WAGSTAFF et al. 1982 ; HOLLINGSWORTH and BYERS 1989 ; ROCKMILL and ROEDER 1990 ; HOLLINGSWORTH et al. 1995 ). In addition, mutations that reduce interhomologue exchange and/or pairing of homologous chromosomes increase spore viability in haploid disomic strains undergoing spo13 meiosis (WAGSTAFF et al. 1985 ; HOLLINGSWORTH and BYERS 1989 ; ROCKMILL and ROEDER 1990 ).

To monitor the segregation of the chromosome VIII pair in DT71, one homologue is marked with a TRP1 gene integrated just 3' of the GPA1 locus. The other homologue is marked with an ARG4 gene at the equivalent location. The GPA1 locus is ~2 cM from CEN 8 (MIYAJIMA et al. 1987 ; FUJIMURA 1989 ). The frequency of reductional, equational, and aberrant segregations was determined in each strain by dyad dissection. The resulting spore colonies were tested for the centromere-linked ARG4 and TRP1 markers. The segregation pattern of the chromosome VIII pair in DT71 is 60.2% equational, 14.4% reductional, and 25.4% aberrant (Table 2).

Meiotic unequal SCR assay:
One of the chromosome VIII homologues carries a tandem pair of arg4::ura3 segments separated by the CUP1 gene (SCR recombination construct, Fig 1). The arg4::URA3 gene, from which the segments were derived, was created by removing the ARG4 coding region and replacing it with that of URA3. The DNA sequences corresponding to the well-characterized ARG4 hotspot were retained, but the activity in the construct used has not been tested. Unequal exchange or nonreciprocal gap repair (which may or may not be accompanied by exchange) between the arg4::ura3 segments on sister chromatids can generate a functional arg4::URA3 gene and duplicate the intervening CUP1 gene (Fig 1). The level of copper resistance is sensitive to the copy number of CUP1 (HAMER et al. 1985 ). Intrachromatid events that generate arg4::URA3 would not duplicate the CUP1 gene. The unequal SCR recombinant is dominant, eliminating any significant contribution of chromosome segregation pattern in this initial analysis. Southern analysis confirmed that our SCR recombination construct functioned as expected (data not shown). There is homology to the SCR construct on the homologue (HH construct), but interhomologue recombination events cannot generate an intact arg4::URA3 gene (Fig 1). Mutants elevated for meiotic unequal SCR were identified using the pick-and-patch plate assay described in detail in MATERIALS AND METHODS (Fig 2).

The frequency of meiotic unequal SCR in each strain was quantified by plating aliquots from liquid sporulation medium onto medium lacking uracil, arginine, and tryptophan and containing 240 µM CuSO₄ · 5H₂O. Viable titer was determined by plating on rich (YEPD) medium. Addition of 5-FOA to the pregrowth regimen (see MATERIALS AND METHODS) eliminated any significant contribution of mitotically generated Ura⁺ cells (data not shown). Sporulation frequency was determined by microscopic examination of liquid sporulation cultures.

In the wild-type haploid disomic strain (DT71), we assume that the majority of the meiotic recombination events occur between homologues, resulting in a characteristic frequency of meiotic unequal SCR, chromosome segregation pattern, and spore viability in spo13 meiosis (Table 2). In contrast, we expected that a mutant defective in meiotic recombination partner choice would increase meiotic unequal SCR at the expense of interhomologue recombination, resulting in a change in chromosome disjunction in favor of equational segregation and increased spore viability. The red1::ADE2 mutant illustrates the spectrum of phenotypes exhibited by a mutation affecting interhomologue exchange, pairing, synapsis, and meiotic recombination partner choice in spo13 meiosis (ROCKMILL and ROEDER 1990 ; HOLLINGSWORTH et al. 1995 ; Table 2; see below).

A red1 mutant is increased for meiotic unequal SCR:
The hypothesis that meiotic sister-chromatid cohesion suppresses meiotic sister-chromatid exchanges suggests that disruption of sister-chromatid cohesion will result in an increase in meiotic SCR. The product of the RED1 gene is required for meiotic sister-chromatid cohesion (SMITH and ROEDER 1997 ; BAILIS and ROEDER 1998 ). We compared the frequency of meiotic unequal SCR in a red1::ADE2 mutant (ROCKMILL and ROEDER 1990 ) with that in RED1 strains. The red1::ADE2 mutant is increased 3.6-fold for meiotic unequal SCR, and the frequency of equational segregation is increased to 95% at the expense of the reductional and aberrant classes. In addition, sporulation frequency (P < 0.001) and spore viability (P < 0.001) are increased (Table 2). We used the red1::ADE2 mutant as a positive control in the screen for mutants with a comparable elevation in meiotic unequal SCR.

The frequency of meiotic unequal SCR in a monosomic (homologue) strain carrying the SCR construct represents the maximum detectable frequency in this system:
The frequency of recombination between duplicated HIS4 sequences is ~10-fold higher in haploid meiosis than it is in the same construct in diploid meiosis (JACKSON and FINK 1985 ; WAGSTAFF et al. 1985 ). This observation was corroborated in a study that compared intersister and ectopic exchanges in isogenic diploid and haploid strains (LOIDL and NAIRZ 1997 ). In addition, meiosis-specific DSBs occur at wild-type levels and are processed efficiently in spo13 haploids (DE MASSY et al. 1994 ; GILBERTSON and STAHL 1994 ). These results imply that recombination between sister chromatids is suppressed in a diploid. In haploid meiosis, however, chromatids with DSBs are able to use the homology available on the sister chromatid for recombinational repair. Similarly, the frequency of meiotic unequal SCR in the monosomic (homologue) haploid strain is increased 5.8-fold compared to that of the disomic haploid strain (DT71, Table 2). We assume that this is the maximum frequency of meiotic unequal SCR we can expect in this system. The difference in the frequency of meiotic unequal SCR in our monosomic strain and the frequency of intersister recombination reported by others in haploid meiosis is likely attributable to differences in the construct used to monitor meiotic SCR.

msc mutants define three classes:
DT71 was mutagenized by integrative transformation with a transposon-mutagenized yeast genomic library carrying random TN3::LEU2 insertions (BURNS et al. 1994 ). We screened 53,523 colonies for an increase in meiotic unequal SCR (for details see MATERIALS AND METHODS). The putative mutants were then screened by dyad dissection for an alteration in the segregation pattern of the chromosome VIII disome. For candidates that satisfied both criteria, DNA flanking the transposon insertion was recovered and sequenced. The locus of transposon insertion was determined by reference to the Yeast Genome Database (see MATERIALS AND METHODS).

To ensure that they exhibited phenotypes relevant to meiosis, the msc (meiotic sister chromatid recombination) mutants were quantified in spo13 meiosis with respect to meiotic unequal SCR frequency, disome segregation pattern, sporulation frequency, and spore viability (see above and MATERIALS AND METHODS). In addition, outcrossing the mutants to a SIR3 SPO13 strain revealed a class that conferred a dominant meiotic lethal phenotype peculiar to our strain background (see class II below). Analysis of the msc mutants according to these criteria defines three classes (Table 2 and Table 3).

Class I:
Mutants in class I are increased in meiotic unequal SCR, and they are increased in equational segregation at the expense of the reductional and aberrant classes (Table 2 and Table 3). A total of 23 mutants fall into this class. These mutations identify alleles of RED1, MEK1, RAD24, MEC3, and MSC6. In the remainder, the phenotype was unlinked to the locus of transposon insertion; these mutants are denoted in Table 2 by strain number (see footnote f) and were not pursued further. Our observation that mutations in several of the meiotic recombination checkpoint genes result in an elevation of meiotic unequal SCR suggests that these genes encode functions required for proper meiotic recombination partner choice.

The nine red1::TN alleles represent eight different insertion positions spanning the open reading frame. Most of the red1::TN mutants exhibit increases in sporulation frequency (P < 0.001) and spore viability (P < 0.001) in spo13 meiosis (Table 2). The nine rad24::TN alleles represent individual insertions spanning the open reading frame. All the rad24::TN insertion alleles have a frequency of meiotic unequal SCR that is higher than the twofold increase observed in the rad24 mutant. Most notably, the insertion at position +625 exhibits a sixfold increase (Table 2).

Single insertions in MEK1 and MEC3 were identified, and linkage of the msc phenotype to the transposon insertion was confirmed. In the mek1::TN+944 mutant, the sporulation frequency is not significantly different from that observed in the majority of red1::TN alleles, but the spore viability approximated that of DT71 (Table 2).

The HOP1 gene is in the same epistasis group as RED1 and MEK1 (ROCKMILL and ROEDER 1990 , ROCKMILL and ROEDER 1991 ). In a reconstruction experiment, a hop1 mutant was not elevated for meiotic unequal SCR, but it did favor equational chromosome segregation and significantly increased sporulation and spore viability (Table 2).

No alleles of RAD17 or MEC1, both of which participate in the mitotic DNA damage and meiotic recombination checkpoints, were identified. In a reconstruction experiment, the phenotype of a rad17 mutant was not significantly different from that of the rad24 mutant (Table 2). Failure to identify any new alleles of RAD17 may imply that the screen was not performed to saturation. MEC1/ESR1 is an essential gene, and the mec1-1 allele is the only viable mutant isolated to date (KATO and OGAWA 1994 ; WEINERT et al. 1994 ). It was subsequently shown that the viability of mec1-1 mutants required an additional mutation in the SML1 gene which results in an increase in dNTP pools (ZHAO et al. 1998 ). We were able to introduce the mec1-1 mutation into our strain, assesed by sensitivity to the DNA-damaging agent methyl methanesulfonate (MMS), suggesting that our strain also carries a mutation in SML1. A mec1-1 mutant is not elevated in the meiotic unequal SCR in our system. However, the spore inviability conferred by this mutation was not rescued by the spo13 mutation (Table 2). Thus, in mec1-1 mutants, any increase in meiotic unequal SCR may have been obscured by spore inviability.

The observed sporulation frequencies and spore viabilities in rad17, mec3::TN+1152, and most alleles of rad24::TN were similar to those of the DT71 strain. The rad24 ::TN+509 allele is exceptional and resembled a red1 mutant in these respects (Table 2).

The strain carrying the yor354c::TN+1608 (mcs6) mutation was not MMS sensitive (data not shown), suggesting that this gene does not function in the DNA damage checkpoint. Expression of the MSC6 gene was not induced in the large-scale study of the transcriptional program of sporulation described in CHU et al. 1998 .

The product of the SPO11 gene catalyzes meiosis-specific DSBs (BERGERAT et al. 1997 ; KEENEY et al. 1997 ). The lack of meiotic induction of SCR events in both spo11 red1::ADE2 and spo11 rad24 double mutants (Table 2) indicates that the meiotic unequal SCR events in the single red1 and rad24 mutants are initiated by meiosis-specific DSBs.

red1 rad24 epistasis:
A red1::ADE2 rad24 double mutant has a meiotic unequal SCR indistinguishable from that of the rad24 mutant, a reduction in sporulation frequency compared to the rad24 mutant (P < 0.01), and a spore viability that is intermediate to the viabilities of the component single mutants (Table 2). The observation that meiotic unequal SCR in the red1::ADE2 rad24 double mutant is not significantly different from that in the rad24 single mutant indicates that RAD24 is required for the elevated levels of meiotic unequal SCR in red1 mutants.

Class II:
Mutants in class II are increased in meiotic unequal SCR, increased in equational segregation at the expense of the reductional and aberrant classes (Table 2 and Table 3), and confer a dominant meiotic lethal phenotype when crossed to a congenic SIR3 SPO13 strain monosomic for chromosome VIII. The spore viability, assessed by tetrad dissection, in each of the corresponding diploids was 1% (data not shown). A total of 11 mutants fall into this class. In two of the mutants, the phenotype was unlinked to the locus of transposon insertion; these mutants are denoted in Table 2 by strain number (see footnote f) and were not pursued further. Single-transposon insertions were identified in INP52, UBR1, BUD3, PET122, and MSC1-MSC3. In one mutant, the transposon insertion position was 246 bp upstream of ELA1, which encodes a yeast elongin A homologue (C. KOTH, personal communication).

Linkage of the class II phenotype to the transposon insertions in INP52, MSC1, and MSC3 was confirmed by transplacement of deletion derivatives of these three genes, respectively, into DT71 and assaying meiotic unequal SCR (see MATERIALS AND METHODS). INP52 encodes an inositol polyphosphate 5-phosphatase that is similar to synaptojannin proteins, which regulate Ca²⁺ levels during neurotransmission (STOLZ et al. 1997 ). MSC1-MSC3 were sequenced as part of the Yeast Genome Project (http://genome-www.stanford.edu/Saccharomyces/) and code for YML128c, YDR205w, and YDR219w, respectively. The gene products presumed to be encoded by MSC1 and MSC3 are not homologous with any proteins in the database. The gene product of MSC2 is a predicted transmembrane protein with homology to S. cerevisiae Cotp, which functions in cobalt ion transport (CONKLIN et al. 1992 ), and to a cation efflux protein in Alcaligenes eutrophus (NIES et al. 1989 ). In a reconstruction experiment, a strain carrying a msc2 mutation was not elevated for meiotic unequal SCR (data not shown), suggesting either that the transposon insertion is not linked to the msc phenotype in this mutant or that this phenotype is specific to the ydr205w::TN+1255 allele.

An allele of UBR1 was isolated in the screen. UBR1 encodes the E3 ubiquitin protein ligase, which associates with the ubiquitin-conjugating enzyme encoded by RAD6 to carry out N-end rule ubiquitin degradation (BARTEL et al. 1990 ). UBR1 is also required for peptide transport into cells (ALAGRAMAM et al. 1995 ). The observation that a ubr1 mutant is not elevated for meiotic unequal SCR suggested either that the transposon insertion is not linked to the msc phenotype in this mutant or that this phenotype is specific to the ubr1::TN+330 allele. It was observed that overexpressing peptides with the N-end rule sequence, recognized by Ubr1p (BARTEL et al. 1990 ), results in a meiotic delay beginning with the appearance of meiotic recombinants in UBR1 strains. However, no delay is observed in ubr1 strains, which proceed through meiosis faster than UBR1 strains (L. BULTÉ, K. MADURA and A. VARSHAVSKY, personal communication). This suggests that partial function of Ubr1p results in the delay in recombinant formation. It may be that, analogously to the case presented above, partial function of Ubr1p is responsible for the msc phenotype of the mutant carrying the ubr1::TN+330 allele.

The msc phenotype of the transposon insertion upstream of ELA1 was complemented by tranformation with a plasmid bearing a wild-type allele of ELA1, indicating that the insertion disrupts expression of ELA1. Experiments to determine linkage of the msc phenotype to the transposon insertions in PET122 and BUD3 are in progress. PET122 encodes a translational activator of cytochrome c oxidase subunit III (KLOECKENER-GRUISSEM et al. 1988 ). The pet122::TN+1593 mutant is able to grow on medium that selects against petite mutants, although it does so more slowly than does DT71. BUD3 encodes a protein required for the axial budding pattern in haploid strains (CHANT et al. 1995 ).

A dmc1 mutant has a class II phenotype:
No alleles of DMC1 were identified in the screen. A dmc1 mutant was shown previously to be elevated for intrachromosomal exchange (BISHOP et al. 1992 ). In a reconstruction experiment, a dmc1 mutant was shown to have a class II msc phenotype (Table 2). This suggests that DMC1 plays a role in directing events to homologous chromatids.

The dominant meiotic lethality of these mutants, when heterozygous, was unexpected, since the dmc1 allele used was shown previously to be recessive for completion of meiosis (BISHOP et al. 1992 ). In addition, a dominant meiotic lethal phenotype has not been reported for mutations in any of the other previously identifed genes in this class (KLOECKENER-GRUISSEM et al. 1988 ; BARTEL et al. 1990 ; CHANT et al. 1995 ; STOLZ et al. 1997 ; C. KOTH, personal communication). Thus, this phenotype apears to be peculiar to our strain background (see DISCUSSION). Since the meiotic phenotypes of the mutants in class II resemble those of dmc1, we suggest that the genes identified by these mutations function in or are regulators of the DMC1-promoted interhomologue exchange pathway.

RED1 and DMC1 act independently in partner choice:
A red1::ADE2 dmc1 double mutant exhibits an additive increase in meiotic unequal SCR, the sporulation frequency approximates that in the single red1::ADE2 mutant, and the spore viability is intermediate to those of the single mutants (Table 2). The frequency of meiotic unequal SCR in the red1::ADE2 dmc1 double mutant is identical to that in the homologue strain, suggesting that meiotic recombination events in this background occur predominantly between sister chromatids. This result corroborates the observation of SCHWACHA and KLECKNER 1997 that red1 dmc1 mutants produce only intersister recombination intermediates in meiosis. The additive increase in meiotic unequal SCR in the double mutant suggests that RED1 and DMC1 act independently to bias the repair of meiosis-specific DSBs to homologues.

Class III:
Mutants in class III are increased in meiotic unequal SCR. In contrast to the mutants in classes I and II, they have increased reductional segregation and generally have a spore viability lower than that of DT71 (P < 0.05 to < 0.01, Table 2 and Table 3). In addition, the mutants in this class exhibit mitotic marker loss, which is likely to be caused by mitotic chromosome loss. There are six mutants in this class; the location of the transposon insertion has been determined in five of them. In three of these, the phenotype was unlinked to the locus of transposon insertion; these mutants are denoted in Table 2 by strain number (see footnote f) and were not pursued further. Experiments to determine linkage of the msc phenotype to the transposon insertions in MNR2 and a previously uncharacterized gene, MSC7, are in progress. The protein encoded by MNR2 has 52% identity to the S. cerevisiae aluminum-resistant protein Alr2p over 98 amino acids (DUJON et al. 1994 ). Overexpression of MNR2 overcomes manganese toxicity. MSC7 was sequenced as part of the Yeast Genome Project and codes for YHR039c, which has some similarity to aldehyde dehydrogenases.

The mutants in this class have the phenotype expected for a meiotic hyper-rec mutation in spo13 meiosis. An elevation in meiotic unequal SCR is expected in a mutant with a meiotic hyper-rec phenotype, since both intersister and interhomologue events exhibit meiotic induction. In addition, a meiotic hyper-rec mutant, in which more interhomologue connections would occur, is expected to display an increase in reductional segregation and a decrease in spore viability. This expectation is supported by the phenotype of recombinationless spo11 mutants in spo13 meiosis, which is an increase in equational segregation and a concomitant increase in spore viability (WAGSTAFF et al. 1982 ). The observed increase in the frequency of reductional segregation in the hyper-rec strains compared to that in DT71 in not statistically signficant (P > 0.08), but is likely to be an underestimate because of the spore inviability correlated with interhomologue exchange. In support of this possibility is the observation that spore death in spo13 disomic haploid meiosis is nonrandom (WAGSTAFF et al. 1982 ). Dyads in which neither or both spores survive occur more frequently than predicted by random spore death, indicating that the majority of spore inviability results from events that are lethal to both spore products in a given meiosis. Class III mutants display an excess of dyads with two inviable spores compared to DT71 (data not shown), supporting the proposal that these mutants are increased for interhomologue exchange.

Meiotic unequal SCR in diploid spo13 strains:
To confirm that the increase in meiotic unequal SCR in the msc mutants is not specific to disomic haploids, the frequency of meiotic unequal SCR was determined in wild-type, red1, mek1, rad24, and mec3 derivatives of a spo13 diploid strain isogenic to DT71. Meiotic unequal SCR was elevated in all mutants compared to the wild type, indicating that the increase in meiotic unequal SCR in the mutant backgrounds is not specific to haploid meiosis (Table 4).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Using a strain designed specifically to detect mutants exhibiting an increase in meiotic unequal SCR, we conducted a screen for components of the machinery that directs meiotic exchange events to homologous chromatids. This approach has identified 38 msc mutants comprising three phenotypic classes. Class I mutants identified genes known and likely to be required for the meiotic recombination checkpoint, class II mutants have a phenotype similar to a dmc1 mutant, and class III mutants are putative meiotic hyper-rec mutants.

Class I:
Genes involved in the meiotic recombination checkpoint also play a role in meiotic recombination partner choice. The meiosis-specific genes RED1 and MEK1 and the DNA damage checkpoint genes RAD24</