Genetic and Physiological Aspects

It has long been known that diploid strains of yeast are more resistant to 7-rays than haploid cells, and that this is in part due to heterozygosity at the mating type (MAT) locus. It is shown here that the genetic control exerted by the MAT genes on DNA repair involves the a1 and a2 genes, in a R M E l independent way. In rad18 diploids, affected in the error-prone repair, the a/a effects are of a very large amplitude, after both UV and 7-rays, and also depends on a1 and a2. The coexpression of a and a in rad18 haploids suppresses the sensitivity of a subpopulation corresponding to the GS phase cells. Related to this, the coexpression of a and a in RAD+ haploids depresses UV-induced mutagenesis in GP cells. For srs2 null diploids, also affected in the error-prone repair pathway, we show that their GI UV sensitivity, likely due to lethal recombination events, is partly suppressed by MAT homozygosity. Taken together, these results led to the proposal that al-a2 promotes a channeling of some DNA structures from the mutagenic into the recombinational repair process. E ARLY work demonstrated that diploid yeast cells are more resistant to the lethal effects of radiations, especially after y-rays, than haploid cells (LATARJET and EPHRUSSI 1949) and that part of this diploid resistance is dependent upon heterozygosity at the mating-type locus MAT (MORTIMER 1958; LASKOWSKI 1960). The experiments reported here concern the control by the MAT genes of this ala-directed diploid repair. The ploidy effects are currently interpreted as a result of the availability of homologous chromosomes allowing recombinational repair to take place in G1 diploids as opposed to G1 haploids. Indeed, a number of physiological properties conferred by mutations in the different DNA repair genes-which define three repair pathways as deduced from the study of genetic interactions between their mutations [for reviews see HAYNES and KUNZ (1981), GAME (1983) and FRIEDBERG (1 988)]-favor this interpretation. The ploidy effects are completely abolished by mutations in any of the three RAD51, RAD52 and RAD54 genes belonging to the recombinational repair group of genes (SAEKI, MACHIDA and NAKAI 1980), which are those essential for double-strand break repair [for a review see GAME (1 983)]. On the contrary, mutations affecting either the xcision or the error-prone repair mechanism do not eliminate the ploidy effects. These are particularly large in mutants of some of the errorprone repair genes. The rev3 mutants, deficient for the presumptive error-prone DNA polymerase (MORRISON et al. 1989), exhibit a considerable G1 UV sensitivity when haploid, whereas, if diploid and heterozygous for the mating-type, their esistance is comGenetics 133: 489-498 (March, 1993) parable to that of wild-type diploids (LEMONTT 1971; LAWRENCE and CHRISTENSEN 1976; MCKEE and LAWRENCE 1979a,b, 1980; CASSIER-CHAUVAT and MousTACCHI 1988). Haploid mutants of the RAD18 gene, which codes for a potential DNA binding protein (CHANET, MAGANA-SCHWENCKE and FABRE 1988; JONES, WEBER and PRAKASH 1988) involved in mutagenic repair [for a review see LAWRENCE (1982) and CASSIER-CHAUVAT and FABRE (1 99 l)], are very sensitive to UV and y-irradiation in both GI and G2 phases, as opposed to rad18 homozygous diploids which, when heterozygous at the MAT locus, present only a slight sensitivity as compared to wild-type diploids (BORAM and ROMAN 1976; SAEKI, MACHIDA and NAKAI 1980). A significant diploid repair has also been reported in MAT heterozygous rad6 homozygous mutants treated by y-rays (SAEKI, MACHIDA and NAKAI 1980). This diploid effect is much smaller than that observed for rev3 and rad18 cells, although rad6 haploids show a sensitivity to the lethal effects of radiations comparable to that of rad18 mutants. The RAD6 gene, which codes for an ubiquitin-conjugating enzyme (JENTSCH, MCGRAWTH and VARSHAVSKY 1987), is also involved in mutagenic repair (LAWRENCE and CHRISTENSEN 1976). Finally, mitotic recombination is more efficient in heterozygous MATalMATa diploids than in homozygous MATalMATa or MATal MATa RAD" cells, both spontaneously (ESPOSITO and WAGSTAFF 1981) and after UV induction (FRIIS and ROMAN 1968; HOPPER, KIRSCH and HALL 1975). Taken together, these physiological data imply that diploid repair, modulated by the MAT locus, requires recombinational functions which permit some repair 490 M. Heude and F. Fabre i n mutants blocked in the error-prone repair pathway. An exception to this restoration is the case of srs2 null mutations. SRS2 is a gene involved in error-prone repair (ABOUSSEKHRA et al. 1989) which codes for a DNA helicase (ABOUSSEKHRA et al. 1989; H. KLEIN personal communication). The SRS2 deletion renders G1 cells UV sensitive, and srs2 homozygous diploids are not more resistant than haploids. This GI sensitivi ty is most likely due to lethal recombination events since it is suppressed by mutations in the RADS1 gene (ABOUSSEKHRA et al. 1992). The information present at the MAT locus defines the three different cell types, i.e., a, a and a la cells. Two genes constitute each of the MAT alleles, a 1 and a 2 for MATa, a 1 and a2 for MATa (ASTELL et al. 1981; KLAR et al. 1981; NASMYTH et al. 1981). The products of a l , a1 and a2 orchestrate the expression of a battery of target genes, especially controlling cell specialization in the life cycle, namely mating, meiosis and sporulation [for reviews see HERSKOWITZ and OSHIMA (1981), NASMYTH (1 982), NASMYTH and SHORE (1987) and HERSKOWITZ (1988)l. The a1 encoded DNA-binding protein operates in haploids to activate the transcription of a-specific genes required for their mating. Its transcription is lowered by at least two orders of magnitude in MATalMATa heterozygotes (KLAR et al. 198 1; NASMYTH et al. 1981). The a2 product of MATa is also a DNA-binding protein, which operates as a negative regulator of transcription. It acts in two different states: as an homodimer (SAUER, SMITH and JOHNSON 1988) in both a haploids and a l a diploids, to repress the aspecific genes required for the mating of a cells, or as an heterodimer with the a1 product (DRANGINIS 1990), in a/a diploids, to repress the transcription of haploid specific genes. Among these is the R M E l gene coding for a zinc finger protein (COVITZ, HERSKOWITZ and MITCHELL 199 1) acting as a negative regulator of meiosis and sporulation (MITCHELL and HERSKOWITZ 1986). The al-a2 repressional activities render diploids heterozygous at the MAT locus unable to mate, but able to carry out meiosis and sporulation in suitable nutritive conditions. Mutations in R M E l partly relieve meiosis completion from MAT control (KASSIR and SIMCHEN 1976). No regulatory role has been shown for the a1 product alone and no function has been assigned to a 2 in Saccharomyces cerevisiae (TATCHELL, NASMYTH and HALL 1981), although a2 appears to affect the extracellular production of glucoamylase in Saccharomyces diastaticus (YAMASHITA, TAKANO and FUKUI 1985). In this work, we questioned the means by which the MAT alleles exert their control on repair and show that the ala-dependent component of diploid repair is mediated by the a1 and a2 alleles. By extending the previous results of HOPPER, KIRSCH and HALL (1975), we confirm that the R M E l gene, the only al-a2 target already characterized as negative regulator for diploid specific functions, is not involved in this al-a2 mediated repair. We also focussed our attention towards the a/a effects on mutagenic repair in order to test the hypothesis that a l a favors recombinogenic versus mutagenic repair of a common substrate. We demonstrate that MAT heterozygosity is essential for the substantial diploid recovery of radioresistance observed in rad18 mutants. Using haploids transformed with plasmids bearing one or the other of the MAT copy, we showed that the presence of both a and a, affects also DNA metabolism when cells are in G2 mitotic phase. Indeed, the UV and y-ray sensitivity of rad18 G2 haploids is functionally suppressed by the concomitant expression of a and a. Furthermore, UVinduced mutagenesis in RAD+ haploids was found to be depressed by the coexpression of a and a, if cells were in GP during the treatment, as expected if some potentially mutagenic lesions were channeled to recombination. We show also that the G1 sensitivity conferred by deletions in the SRS2 gene of the mutagenic pathway is partially ala-dependent, which suggests that the MATalMATa genotype favors recombination events between homologous chromosomes that are abortive in absence of the Srs2 helicase activity. MATERIALS AND METHODS Media: Complete glucose (YEPD), sporulation, minimal (MM) and the various omission media were as in SHERMAN, FINK and LAWRENCE (1974). Plasmids: The YCp50-MATa plasmid was pJM9 kindly provided by G. SIMCHEN. The 4.2-kb Hind111 fragment containing MATa is inserted in the corresponding site of the YCp50 vector. The YCp50-MATa was constructed by cloning the 4.4-kb EcoRI fragment carrying MATa from a YEpl3-MATa plasmid into the corresponding sites of YCp50 plasmid. The YEp13-MATa plasmid was kindly provided by R. SCHIESTL and U. WINTERSBERGER. Strains and strain constructions: Strains used in this study are listed in Table 1. The matmutations were derived from the {and XJ104-25a strains. The construction of the mata2strains was initiated by crossing the XJ104-25a strain with an RAD+ ura3-52 strain. The sirl-1 mutation allows efficient mating of the XJ104-25a cells. The resulting ura3 heterozygous diploids were then exposed to a sublethal dose of y-rays (50 Gy) in order to induce ura3 homozygous recombinants for subsequent transformation by the YCp50MATa plasmid. The transformed diploids were sporulated, and a2ura3 spores with or without the plasmid were isolated. Those spores carrying the YCp50-MATa were then used for strain constructions through classical genetic procedures. The presence of the a2mutation was ascertained at each step by verifying the non-mating phenotype of haploids and the sporulation deficiency of diploids, both mutant phenotypes being reversed by the YCp5O-MATa plasmid. Due to the ura3-52 genotype of the initial a l strain, transformation with the YCp50-MATa plasmid allowed easy construction of a l strains with new genotypes. The a l genotype was ascertained in a l / a diploids by their inability to sporulate and by their a-mater phenotype in absence of the YCp50-MATa plasmid. Homozygous MAT diploids were obtained from the MAT heterozygous strains ala-control of DNA repair