Mol Gen Genet (1996) 252:530-538

© Springer-Verlag 1996

Montserrat Elias-Arnanz • Antoine A. Firmenieh Paul Berg

Saccharomycescerevisiae mutants defective in plasmid.chromosome recombination

Received: 20 March 1996/Accepted: 31 May 1996

Abstract We have studied the recombinational repair of a double-strand break (DSB) in a plasmid-borne ade2:: HO-site by an intact ade2 allele following the induction of a galactose-inducible GAL-HO gene. If GAL-HO expression is not attenuated by the presence of a low level of glucose in the galactose medium, deleterious effects are observed. Our comparison of the effects of several tad mutations on the relative efficiencies of DSB repair at both the ade2 : : HO-site and at the chromosomal M A T locus indicate that the two processes share common functions. Not surprisingly, most of the recombination-defective mutants found using our assay are alleles of genes in the RAD52 epistasis group. The recombination and repair deficiencies vary among the different mutant groups and also among mutants within a group. In general, there is a correlation between the extents of the recombination and repair defects. Our screen also turned up a novel rfal allele with a pronounced deficiency in DSB repair and recombination and a srs2 mutation which causes only a mild defect. Key words Recombination assay. Double-strand break repair • tad alleles

Introduction

The genetic control of homologous recombination is best understood in Escherichia coli, where at least 25 genes have been implicated in the process and where many of the proteins and the reactions have been characterized (for review, see Kowalczykowski et al.

Communicated by R. Devoret M. Elias-Arnanz - A. A. Finnenich - P. Berg (N~) Department of Biochemistry, Beckman Center for Molecular and Genetic Medicine, B062, Stanford University School of Medicine, Stanford, CA 94305-5425, USA

1994). By contrast, our knowledge of the participants in, and mechanisms of recombination in eukaryotes is less advanced, much of what is known having been learned from studies of both mitotic and meiotic recombination in Saccharomyces cerevisiae and mutants defective in these processes (reviewed in Petes et al. 1991). The most thoroughly studied class of recombination-defective mutants comprise the RAD52 epistasis group, virtually all members of which were identified on the basis of their sensitivity to double-strand breaks (DSBs) induced by ionizing radiation (see Petes et al. 1991). Indeed, the close association of a deficiency in repair of DSBs with a defect in recombination suggests that these two processes proceed via common reactions (Resnick and Martin 1976; Malone and Esposito 1980; Weiffenbach and Haber 1981; Szostak et al. 1983). Mutants variably defective in mitotic recombination have also been isolated by more direct screens, such REC genes having been identified by following recombination in strains disomic for chromosome VIII (Rodarte-Ramon and Mortimer 1972) or chromosome VII (Esposito et al. 1984; Esposito and Brown 1990). More recently, Hoekstra et al. (1991) identified a new yeast gene (HRR5) involved in the repair of DSBs; its product appears to be a protein kinase associated with cell growth and DNA metabolism. Considering the complexity of the reactions and regulatory processes required for recombination and repair, we presumed that the mutants already identified constitute an incomplete collection. Accordingly, we initiated a search for additional S. cerevisiae mutants that fail to carry out efficient mitotic recombination. Our screen relied on recombination between two ade2 alleles, one present on a centromeric plasmid and the other at its normal chromosomal locus. The plasmidborne ade2 allele contains an insertion of the HO endonuctease recognition site. High frequency recombination is initiated by cleavage of the HO site following induction of a GAL-HO endonuclease gene with galactose. Repair of the plasmid's broken ade2 allele by

531 r e c o m b i n a t i o n with the c h r o m o s o m a l a d e 2 - A 1 allele eliminates the inserted H O site, t h e r e b y yielding a funct i o n a l A D E 2 allele. I n the w i l d - t y p e strain, this event occurs so frequently t h a t virtually all of the u n i f o r m l y red colonies characteristic of a d e 2 m u t a n t s d e v e l o p white papillae as the result of c r e a t i o n o f a f u n c t i o n a l A D E 2 gene. M u t a n t s t h a t are u n a b l e to repair the b r o k e n a d e 2 allele b y r e c o m b i n a t i o n fail to p r o d u c e papillated colonies, a readily detectable p h e n o t y p e . A d e s c r i p t i o n of the screen for such r e c o m b i n a t i o n m u t a n t s , w h i c h led to the isolation a n d c h a r a c t e r i z a t i o n o f a n o v e l r f a l allele, c a n be f o u n d in F i r m e n i c h et al. (1995). T h e p r e s e n t p a p e r has t w o principal aims. O n e is to p r e s e n t s o m e p r e v i o u s l y u n r e p o r t e d features of o u r assay t h a t are relevant for u n d e r s t a n d i n g the basis for o u r screen. T h e s e c o n d is to characterize the m u t a n t s f o u n d in o u r screen m o r e fully so t h a t their existence a n d availability are k n o w n . D e s p i t e the fact t h a t m o s t o f the alleles b e l o n g to p r e v i o u s l y identified R A D genes, o u r collection o f new alleles s h o u l d be o f c o n s i d e r a b l e interest. F o r instance, m a p p i n g the m u t a t i o n s in several different alleles o f the s a m e gene c o u l d help t o identify f u n c t i o n a l d o m a i n s ; also, m o n i t o r i n g the efficiency of r e c o m b i n a t i o n using alleles displaying different degrees o f i m p a i r m e n t c o u l d aid in assessing the roles o f the c o r r e s p o n d i n g genes.

Materials and methods

Siede, respectively. The rad51, rad54, rad55 and rad57 disruption plasmids were kindly provided by Rodney Rothstein. YME14 contains a deletion of a 2.85 kb PstI fragment (80% of the coding region) in the chromosomal SRS2 gene. Strains derived from W303 with disruptions in rad genes were obtained from Rodney Rothstein and are, therefore, not listed in Table t.

Plasmids Plasmids constructed specifically for our assay are described in Firmenich et al. (1995). Plasmids containing wild-type RAD genes (used to classify the mutants) were obtained from different sources: pNF1000 (Yep-RADI) and pNF101 (Yep-RADIO) were gifts from Errol Friedberg; YEp13-RAD52 was provided by David Schild; YEp-RAD51, YEp13-RAD54, YEp13-RAD55 and YEP13-RAD57 were obtained from the Yeast Genetic Stock Center (YGSC, Berkeley); pRS12 (Yep13-RAD9) was a gift from Robert Schiestl.

Media and culture conditions YPAD (2% dextrose, 2% peptone, 2% yeast extract, 40 gg/ml adenine) was used to grow yeast strains when marker selection was not required. Synthetic dextrose complete, omission media and minimal medium were prepared as described by Sherman et al. (1986). Recombination was induced on solid minimal medium containing 0.5% glucose, 1% galactose, 10 gg/ml adenine and 20 gg/ml histidine. Survival following continuous generation of HO-induced DSBs was scored on minimal plates containing 2% galactose, 10 gg/ml adenine and 20 gg/ml histidine. Mating, sporulation and tetrad dissection were performed as previously described (Sherman et aI. 1986). Yeast cells were transformed by electroporation (Becker and Guarente 1.991) or by the PEG lithium acetate method (Schiestl and Gietz 1989; Firmenich and Redding 1993).

Strains Isolation and classification of recombination mutants The S. cerevisiae strains used in this work are listed in Table 1. All strains are isogenic with the S288C-derived strain JMll6 (kindly provided by John McCusker). The his3-A allele contains an internal deletion of a 187-bp HindIII fragment (Scherer and Davis 1979). The ade2-A1 allele was constructed by removing a 219 bp HpaI-StuI fragment from the gene. Both deletion alleles were introduced into JM116 to construct the parental strain YME2. In strain YME3, the HO recognition site at the M A T locus was removed using plasmid pJHll3, kindly provided by James Haber. Several disrupted RAD genes were introduced into YME2 and YME3. The tad52 and radl disruption plasmids were gifts from David Schild and Wolfram

TabLe 1 S. cerevisiae strains used in this study

The procedures used to isolate and classify our collection of 78 recombination mutants have been described (Firmenich et al. 1995). Briefly, after mutagenesis, colonies that remained uniformly red, indicative of a defect in recombinational repair of the disrupted ade2 : : HO allele, were picked and screened to eliminate those that were not defective in recombination. Isolates showing a defect in DNA repair were crossed to a set of rad strains (radl, rad6, rad9, radlO, radlS, tad50, tad51, rad52, tad54, tad55 and rad57) as well as transformed with plasmids containing wild-type RAD genes. Mutants that could not be assigned to any of the rad tester genes were

Strain ~

Genotype

Strain"

Genotype

JMl16 YME2 YME3 YME4 YME5 YME6 YME7 YME8 YME13

M A T a ura3-52 Ieu2-A1 M A T a ura3-52 Ieu2-A1 ade2-A1 his3-A matA ura3-52 leu2-A1 ade2-A1 his3-A MATc~ ura3-52 leu2-A1 ade2-A1 his3-A M A T a tad52:: LEU2 M A Tc~ rad52 : : LEU2 M A T a radl :: URA3 M A T e radl :: URA3 matA tad52 :: LEU2

YME14 YME15 YME16 YME17 YME18 YME19 YME20 YME21 YME22 YME23 YME24

M A T a srs2-A matA s~s2-A M A T a rad51 :: LEU2 matA rad51 :: LEU2 M A T a rad54: : LEU2 matA rad54 : : LEU2 M A Ta tad55 :: LEU2 matA tad55 :: LEU2 M A Ta rad57 : : L E U 2 matA rad57 : : LEU2 matA tad1 :: URA3

"Strains YME5-YME25 are isogenic to YME2 except for the markers indicated

532 crossed to one another to identify the number of complementation groups.

Determination of UV and X-ray sensitivities Quantitative and semiquantitative measurements of the radiation sensitivities were performed as described (Fimlenich et al. 1995). Briefly,mid-log cultures were sonicated, serially diluted in water and aliquots were plated on YPAD. Sensitivityto UV was determined by counting the number of survivors following exposure to 0d20 J/m 2. Sensitivity to X-rays was determined after treatment of the plates with 0-40krads at a dose rate of 1193 rads/min. For semiquantitative estimates of radiation sensitivity, 5-gl drops containing 5-9 x 105 cells were spotted on YPAD plates. The plates were allowed to dry for 5 min prior to irradiation with UV or X-rays. Plates were incubated at 30°C for 2 days.

Recombination capability for HO-induced events To compare the recombination proficienciesof the mutants and the parental strain, cells containing pAF30 or pAF35 were plated on the induction medium. After incubation for 5 days, about 40 colonies were pooled (this averages out differencesthat would be expected to occur among diflbrent colonies)and serial dilutions were plated both in complete medium (SD) and medium lacking adenine (SD minus Ade). The number of successful recombination events at the ade2 locus is expressed as the fraction of ADE + cells (colonies on SD minus Ade) relative to the total number of cells (colonies on SD). Survival on galactose under conditions where HO-induced DSBs are continuously produced, is an additional assay for measuring recombination/repair deficiendes quantitatively. Cells with one or two HO target sites (either at the MAT locus, at the ade2:: HO-site or both) were transformed with a GAL-HOendonuclease-expressing plasmid and plated on selective media containing either 2% galactose or 2% glucose. The percentage of cells that survived and produced colonies on the galactose medium compared to the number of colonies on the glucose medium measures the ability of cells to repair the DSBs at the MAT locus, at the plasmid or both.

Results and discussion Characterization of the recombination assay O u r screen for recombination mutants relied on the use of the site-specific H O endonuclease to increase the recombination frequency above the spontaneous levels (Firmenich et al. 1995). T h e two basic c o m p o n e n t s in our assay were a plasmid-borne ade2 allele susceptible to cleavage by H O endonuclease (ade2:: HO-site) and the HO endonuclease gene expressed from the inducible, strong GALI-IO p r o m o t e r (Jensen and Herskowitz 1984). While trying to develop a color assay for recombination based on the ade2::HO-site/HO endonuclease system, we found that continuous expression of the inducible GAL-HO gene after plating on galactose plates was lethal even for the wild-type strain YME2. The observed lethality originates from a failure of Y M E 2 containing pade2 :: HO-site and pGAL-HO (or pAF30, see Firmenich et al. 1995) to repair the DSB at either of the two H O sites (the H O site at ade2 and

the one at MAT); more to the point, the failure to repair the ade2 :: H O allele makes a major contribution to the lethality (Table 2). Thus, the plating efficiency of Y M E 2 cells carrying pade2 :: HO-site and pGAL-HO on a medium containing galactose is only about 8% c o m p a r e d to 92% where neither p G A L - H O nor pade2 :: HO-site are present, i.e., where there is no HO expression (Table 2, control vs. pade2 :: HO-site + pGAL-HO). The viability was increased almost six-fold (to 45%) when maintenance of pade2 :: HO-site was not selected for (that is, in the same medium containing uracil). This indicates that the failure to repair the DSB generated at the plasmid's H O site contributes significantly to the increased lethality observed on galactose. The remaining two-fold reduction in plating efficiency (from 45% to 90% in the control) can be accounted for by the presence in Y M E 2 cells of a second H O target sequence, the naturally occurring site at the M A T locus. This assumption is supported by the plating efficiency on galactose of Y M E 2 transformed with pGAL-HO alone; the reduction in plating efficiency observed in this case (exclusively due to unrepaired breaks at MAT) was almost identical to that of Y M E 2 transformed with both plasmids, when selection for pade2:: HO-site was not applied. The significant contribution of unrepaired breaks at the plasmid's H O site is confirmed by c o m p a r i s o n of the plating efficiencies of Y M E 3 containing pade2:: HO-site and pGAL-HO in the presence and absence of selection for pade2 :: H O site. As expected for a site-specific endonuclease like H O , the plating efficiency of Y M E 3 (which has no c h r o m o s o m a l H O site) containing pGAL-HO alone was similar to that of the control. In contrast to our results, Bennett et al. (1993) found that induction of a DSB in a dispensable plasmid was lethal in RAD ÷ cells. Because the mechanism of this DSB-induced lethality appears to be subject to genetic control, the differences between our results and those by Bennett et al. (1993) could result from genetic differences between the two strains. W h e n selection for the plasmid containing the H O site was imposed, survival on galactose was strictly dependent on recombinational repair of the b r o k e n ade2 alleles. Colonies that grew on these plates were either white (about 70%) or all red (about 30%). Restriction analysis of plasmid D N A isolated from several colonies of the two kinds showed that the H O site had been eliminated in both cases: as expected, the white colonies contained a wild-type ADE2 allele; the red colonies, on the other hand, contained ade2 m u t a n t alleles that, while repairing the break at the H O site, had also copied the deletion m u t a t i o n from the chromosomal ade2-A1 allele (data not shown). As a consequence of the elimination of the H O site in both cases, the recombinants can grow on galactose. We did not examine whether or not the M A T locus in the surviving colonies retained the H O site. It is possible that the survivors on galactose have eliminated the H O site at

533 Table 2 Plating efficiency on galactose; repair of one versus two H O sites after induction of the GAL-HO gene

Strain

Selection for retention of pade2 : : HO-site

M A T (HO site) YME2

+

matA (no HO site) YME3

+

Control a

pGAL-HO

pade2 : : HO-site + pGAL-HO

92 90 90 89

49 53 89 86

8 45 14 88

"Plating efl]ciency represents the percentage of colonies on galactose relative to glucose plates (see Materials and methods). Values shown are means of three experiments. The control column shows the plating efficiency in the absence of pade2: : HO-site and pGAL-HO, i.e., in the absence of DSBs. The pGAL-HO column shows plating efficiency when double-strand breaks are introduced at M A T alone (in YME2). pade2:: HO-site + pGAL-HO gives the plating efficiency when DSBs are produced at M A T and the ade2 : : HO-site allele (if in YME2) or only at the ade2: : HO-site allele (if in YME3). When present, selection for pGAL-HO was applied; selection for pade2 : : HO-site was as indicated. Alternate names for pade2::HO-site and pGAL-HO used elsewhere (Firmenich et al. 1995) are p M E l l 9 and YCpGAL-HO, respectively. Here we use these two plasmids instead of the combined plasmid pAF30 because they allow independent selection for the ade2 :: HO-site and the GAL-HO genes (for information on plasmid construction, see Firmenich et al. 1995)

M A T by some nonhomologous repair event. In this regard, small deletions surrounding the recognition site for HO endonuclease have been reported (Rudin and Haber 1988). Although the recombinational repair of the ade2::HO-site allele and mating type switching are both initiated by cleavage of the HO site, there are some differences between the two processes. In the recombination assay, the allele containing the HO recognition site is in a plasmid and the repair event is interchromosomal in nature. In normal mating type switching, both alleles (one of them a silent copy) are located on the same chromosome and, therefore, the event is intrachromosomal. Differences of this kind may result in different gene requirements for DNA repair. In this regard, Sugawara et al. (1995) have shown that while recombination between a chromosomal M A T a and a plasmid-borne MATc~-inc copy is RAD51 independent, RADS1 is required if the MATc~inc copy is in the chromosome and M A T e on a plasmid. Therefore, it was important to determine if the RADS1-RAD57 genes, which are required for matingtype switching (Malone and Esposito 1980; Weiffenbach and Haber 1981; Petes et al. 1991; Aboussekhra et al. 1992), are required to the same extent for the repair of the broken ade2 allele. We tested this point by comparing the effect of several rad null alleles on recombinational repair of the plasmid's broken ade2 allele in the MAT-containing (YME2) and MAT-deleted (YME3) strain (Table 3). Quite clearly, the mutations in rad51, rad52, rad54 reduce the proficiency of recombination 100 to 400-fold in YME3. The emciency of recombination is also reduced 50- to 70-fold in strains containing rad55 and rad57 disruption alleles (these measurements were made at 30 ° C, but at lower temperatures these two strains are even more defective; see Hays et al. 1995). These results show that the RAD51RAD57 genes, mutations in which block mating-type

Table 3 Relative efficiencies of several tad null mutations in the presence and absence of cleavage at the M A T locus Gene conversion to ADE2 (relative decrease from wild-type) Mutation

YME2 ( M A T )

YME3 (rnatA )

tad51 :: LEU2 tad52 : : LEU2 rad54 : : LEU2 tad55: :LEU2 tad57: :LEU2 radl :: URA3

5100 8000 5500 320 400 80

100 400 280 55 66 33

All the strains contained pAF30, except for the radl :: URA3 strain, which contained pAF35 (for a description of pAF30 and pAF35 see Firmenich et al. 1995). Gene conversion to ADE2 was measured at 30°C as described in Materials and methods. Gene conversion values for the parental strains YME2 and YME3 are typically about 20%. Values are the means of two experiments

switching, are also needed for the repair of the broken ade2 allele, suggesting that both repair events are mechanistically similar. The relatively lesser effect of the radl disruption allele in the two genetic backgrounds is consistent with the fact that RAD1 is not involved in mating-type switching but is needed in DSB repair when the ends contain regions of nonhomology (Fishman-Lobell and Haber 1992). Not surprisingly, the effect of the rad51-rad57 mutations in the gene conversion assay is more pronounced in YME2 than in YME3. We presume that this effect is indirect and results from the requirements for repair of both the plasmid ade2:: HO-site and the HO target at MA T. It was for this reason that we used YME2 as the parental strain for our mutant search. Thus, mutations that block the repair of breaks at both HO sites (such as the rad51-rad57 group) are more likely to be detected in our assay. On the other hand, if there are functions

534

needed for recombination in the plasmid-chromosome recombination system that are dispensable for matingtype switching, these could be detected regardless of the presence of the HO site at M A T . The lethal effect caused by continuous expression of the HO endonuclease gene was overcome by using a mixture of glucose plus galactose as the induction medium for recombination (see Materials and methods and Firmenich et al. 1995). On plates containing a combination of the two sugars, both the parental YME2 and recombination-deficient strains containing pAF30 form visible colonies at about the same plating etficiency. As cells consume glucose and begin to utilize galactose, YME2 colonies display white papillae emerging from the originally red colonies, while recombinationdeficient strains form uniformly red colonies. Our screen for recombination-deficient mutants used YME2 cells transformed with both pAF30 and pME137, a ptasmid with a his3 allele containing an HO target site (Firmenich et al. 1995). Those mutants that were defective for recombinational repair at both loci and also passed a number of secondary screens were characterized further. Although repair of the broken his3 allele was used to eliminate unwanted mutations leading to the all-red colony phenotype, e.g., mutants in the adenine biosynthetic pathway, we obtained a large number of the colonies that appeared to be deficient for recombination at both loci (between 30-50% of the initially picked colonies). They turned out to be adel2 or ade13 mutations (Dorfman 1969). To our knowledge, this observation constitutes the first evidence that mutations in these two loci confer simultaneous auxotrophy for adenine and histidine. It is well documented that mutations in ade3 causes the same double requirement (Jones and Magasanik 1967; Mazlen t968) but the double mutants ade2 ade3 produce white and not all red colonies. Because adel2 and ade13 mutants cannot grow on hypoxanthine, they could be easily eliminated by including hypoxanthine instead of adenine in the induction plates. All in all, 78 isolates remained after the various screens confirmed their impairment in our assay for gene conversion. Classification and complementation analysis of the 78 mutants The classification of our mutants into complementation groups had to take account of the fact that many genes involved in recombination are also involved in the repair of DNA damage (for review, see Petes et al. 1991). Accordingly, the 78 isolates were tested for sensitivity to UV and X-rays using a qualitative spot assay. The majority of the mutants (63) were sensitive primarily to X-rays, five were extremely sensitive to UV and the remaining 10 were only marginally sensitive to both agents. Complementation analysis with strains containing known rad genes and rescue o f the defective

Table4 Classification of the 78 isolates into complementation ~oups Locus

Number of alleles ~

Allele designation

RAD1 RADIO RAD9 RAD50 RAD51 RAD52 RAD54 RAD55 RAD57 RFA1 SRS2 Unknown

4 1 1 1 19 12 13 7 10 1 1 9

radl-30 to -34 radlO-30 tad9-30 tad50-30 tad51-35 to -53 rad52-30 to -41 rad54-30 to -42 tad55-30 to -37 rad57-30 to -39 rfal-44 srs2~41

"The classification into complementation groups is based on genetic complementation tests with known rad strains and transformation with plasmids containing wild-type RAD genes. One double mutant (tad52, tad54) was found, which contained the alleles we have designated rad52-37 and rad54-33. This explains why the Table includes 79 alleles although the total number of isolates was 78. To assign contiguous numbers to our mutants, avoiding allele numbers that have been previously given by others, our alleles have been numbered starting from 30. Since the allele numbers tad51-31, rad51-32, tad51-33 and tad51-34 had already been taken, the tad51 alleles were numbered from 35 on

phenotypes with wild-type RAD genes led to the classification of the 78 isolates into at least 11 complementation groups. Table 4 shows the distribution of all the isolates into specific complementation groups as well as their allele designation. The majority of the mutations proved to be alleles of genes in the RAD52 epistasis group. Of these, all the complementation groups, with the exception of RADSO, were represented by more than a single allele. The five UV-sensitive strains contained mutations in either RAD1 (four isolates) or RADIO (one isolate). This reinforces the notion proposed by Fishman-Lobell and Haber (1992) that RAD1 and RADIO, although primarily involved in nucleotide excision repair, are also required in double-strand break repair when the ends contain regions of nonhomology. We also identified an allele of RAD9, a gene involved in the control of cellcycle arrest in response to DNA damage (Weinert and Hartwell 1988). All of the above rad alleles behaved as recessives in heterozygous diploids, with the exception of several rad51 alleles: rad51-40, rad51-43 and radS149 behaved as dominant alleles and rad51-37, tad51-39 and rad51-51 behaved as semidominant alleles. A semidominant rad51 allele (tad51-10) has been previously described (Aboussekhra et al. 1992). The Rad51 protein shares some similarities with RecA (Shinohara et al. 1992; Basite et al. 1992; Aboussekhra et al. 1992; Ogawa et al. 1993; Story et al. 1993), for which several mutant alleles with the same semidominant effect have also been found (reviewed in Kowalczykowski et al. 1994). Inasmuch as RecA and RecA-like proteins probably

535 function in vivo as a multimeric filament associated with DNA, this behavior of dominant negative RecA and RecA-like proteins may result from the formation of dysfunctional heteromultimers (with mutant and wild-type protein). Eleven isolates, comprising one radiation-sensitive mutant (mutant 44) and 10 strains with marginal DNA repair phenotypes, were not complemented by any of the known RAD genes and were, therefore, studied further. Phenotypic complementation of the recombination defect of mutant 44 with a yeast genomic library identified the affected gene as RFA1, which encodes the large subunit of the heterotrimeric yeast single-strand DNA binding protein RPA (Heyer et al. 1990; Brill and Stillman 1991). A complete description of the cloning procedure and the characterization of rafl-44, as well as a model for how RFA1 might function in yeast recombination and repair has already been published (Firmenich et al. 1995). Of the strains displaying only slight sensitivity to UV and X-rays, three - isolates 15, 36 and 41 - were subjected to further genetic analysis. When these mutants were backcrossed to YME2, the resulting diploids behaved like wild type, indicating that the corresponding mutations were recessive. Dissection of tetrads from each of the diploids showed that the mutations in isolates 15, 36 and 41 segregate as single genes. Complementation tests indicated that they represent three different complementation groups. So far, we have identified the affected gene in mutant 41 only. A clone that complements the recombination defect of mutant 41 was isolated using a yeast genomic library constructed by Rose et al. (1987). Partial sequencing of the clone revealed identity to SRS2, a gene that encodes a DNA helicase involved in DNA repair (Aboussekhra et al. 1989). Evidence supporting the identification of mutant 41 as affecting SRS2 comes from crosses with YME14 (a strain containing an srs2 deletion allele); such diploids have meiotic defects, e.g. reduced sporulation capacities and reduced spore viability, properties which have been noted previously for srs2/srs2 diploids (Palladino and Klein 1992). Also consistent with the allelism between mutant 41 and SRS2 is the fact that, like other srs2 alleles, mutant 41 was able to suppress the radiation sensitivities of tad6 and rad18 mutants (Schiestl et al. 1990; Palladino and Klein 1992) (data not shown). Although the mutant allele from mutant 41 has not yet been cloned, our observations are entirely consistent with the view that mutant 41 is defective in SRS2 and therefore, we designated the new allele as srs2-41. Attempts to clone the gene affected in mutant 15 have been unsuccessful because this isolate is only slightly defective in forming papillated colonies, thereby complicating the detection of complementation after transformation with a yeast genomic library. The fact that only one allele of some complementation groups was found suggests that we may have not

identified all the functions involved in this particular recombination event. Quite possibly, using hypoxanthine instead of adenine in the medium to screen for putative mutants (see above) would eliminate most of the false positives and, thereby, help to maximize the recovery of rare and novel mutations. The genes affected in nine strains displaying minor DNA recombination and repair phen0types, and representing at least two complementation groups, remain to be identified. It is possible that these mutations occur in genes that have an important role in recombination but are leaky; this was the case for some rad52 alleles (i.e., rad52-30 and tad52-35, see below). Alternatively, the genes affected may have a redundant function rather than a primary role in the repair process. Quantification of the deficiencies in recombination and repair The semi-quantitative spot assay suggested that the 78 mutant strains obtained in our screen were variably defective. More quantitative data on the impairment are summarized in Table 5. We selected strains from each complementation group representative of the varying degrees of impairment and determined their recombination deficiencies in two different but related ways. One relied on the same recombination assay used for their isolation. The parental strain YME2 and 20 mutant strains (representing 12 different complementation groups) containing pAF35 (Firmenich et al. 1995) were plated on induction medium and after 5 days, the fraction of ADE + cells relative to the total number of cells was determined (see Materials and methods). Although this method does not provide recombination frequencies, it is a useful and valid way of quantifying the defects in our mutants relative to one another and to the parental strain in our specific plating assay. The ability to repair HO-induced DSBs was also quantified by determining the fraction of cells surviving on galactose relative to glucose (see Materials and methods). The recombination deficiencies (expressed as the relative decrease with respect to wild type) varied amongst the various mutant groups, from several-fold to more than a thousand-fold for some alleles of members in the RAD52 epistasis group. Mutant 15 and the strains containing the rad50-30 and rad9-30 alleles had the mildest phenotype with only about a five-fold reduction compared to the wild-type strain. The srs2-41 strain and mutant 36 also showed a modest reduction (10-fold and 14-fold, respectively). Mutations in the RAD51, RAD52 and RAD54 genes caused more severe defects, some greater than a thousand-fold. The rfal-44 strain was almost 400-fold less competent for recombination, about as defective as the rad55-37 or tad57-38 strains at 30 ° C. Variability was also observed amongst different alleles within a particular complementation

536 Table 5 Repair deficienciesof strains bearing representative alleles of the differentcomplementationgroups Mutant (strain)

Gene Survival after conversion to exposure to ADE2" 20 kradsb (relative decrease from wild-type)

radI-32 (YME30)~ radl-33 (YME31)* radg-30 (YME32) radSO-30 (YME33) rad51-36 (YAF21) tad51-53 (YME34) rad52-30 (YME35) rad52-31 (YME36) rad52-35 (YAF9) rad52-38 (YAF22) raf54-30 (YME39) rad54-34 (YAF23) rad55-30 (YME40) rad55-37 (YAF37) rad57-30 (YME41) rad57-38 (YAF56) rfal-44 (YAF44) srs2-4l (YME42) # 15 (YME43) # 36 (YME44)

70 21 5.5 4.1 > 1123 2300 16 1800 35 730 3500 > 1818 52 369 84 425 390 10 5 14

1.2 1 9 286 > 1000 860 3 2800 ND 1450 1400 > t000 14 400 43 ND 58 1 1 2

aAll strains containedpAF35 (Firmenichet al. 1995).Gene conversion to ADE2 was measuredas describedin Materialsand methods. For the parental Strain YME2, values of conversion to ADE+ of about 20% are typical bIrradiation with X-rays (20krads) was done immediately after plating. At the X-ray dose used, the wild-typestrain YME2 shows 8% survival cThe radl-32 and tad1-33 mutations,like other RAD1 alleles,caused extreme sensitivityto UV irradiation, being 12800- and 7700-fold more sensitivethan the parental strain, respectively(when exposed to 50 J/m2) ND, not determined

group; for example, rad52-31 and rad52-38 were considerably more defective than rad52-30 and tad52-35. All the strains displaying decreased recombination in the ade2 gene conversion assay were also tess viable on galactose than the parental strain YME2 (data not shown). In general, there was a strong correlation between the inability to survive double-strand breaks induced by cleavage with HO endonuclease and the gene conversion assay. Considering the strong effects of mutations in the RAD52 group (of which RFA1 seems to be a member; see Firmenich et al. 1995), the majority of the DSBs generated in the ade2 system must be processed via the RAD52 recombinational repair pathway. The effect of the srs2 mutation in our system, although mild compared to those in the RAD52 epistasis group, was nevertheless detectable. The SRS2 gene, which is believed to act in a repair pathway in combination with the R A D 6 and RAD18 genes (Petes et al. 1991; Rong et al. 1991; Palladino and Klein 1992), seems to be able to process a low proportion of the breaks produced in our recombination system (either

as such or following processing of the break site to allow them to be recognized by their repair pathway). The effect of the rad9 mutation can be explained by its requirement to arrest cell division in response to DNA damage (Weinert and Hartwell 1988). As shown (see Table 2), failure to repair the HO-induced DSBs will cause cell death. The RAD9-dependent arrest may be necessary to ensure repair of these breaks and cell viability. The selected isolates were also assayed for their sensitivities to UV, X-rays or both. Since the majority of the strains were primarily sensitive to X-rays, Table 5 only includes the value for their X-ray sensitivities. As was the case for their recombination deficiencies, the radiation sensitivities (expressed as fold-decrease with respect to the parental strain YME2) were distributed over a wide range. Moreover, the extent of the repair defects usually correlated with the severity of the recombination defects. Thus, strains that were more sensitive to irradiation were also more severely defective in recombination. The single rad50 allele, rad50-30 (and a disrupted rad50 allele, data not shown), was the exception in that, unlike other mutations in the RAD52 epistasis group, the rad50-30 mutant is almost 300-fold more sensitive to X-ray damage than the wild-type strain but only four-fold less proficient in the ade2 recombination assay. Phenotypic differences between rad50 mutants and other mutants in the same epistasis group have been noticed previously. Thus, during mitotic growth, spontaneous recombination is abolished by rad51 and rad52 mutations whereas mutations in RAD50 exhibit a hyperrecombination phenotype (Malone and Esposito 1980; Malone et al. 1990). Also, in contrast to other members of the RAD52 group, rad50 cells caused a delay but do not prevent switching at M A T (Ivanov et al. 1994). The effect of rad50 mutations in recombinational repair of HO-induced DSBs at M A T is due to a decrease in the 5'-3' degradation at the ends of the break (Ivanov et al. 1994). The small effect caused by the tad50-30 mutation in our gene conversion assay, another case of HO-induced recombination event, may also reflect only a delay in the completion of the process as a result of reduced 5'-to-3' degradation. Analysis at the DNA level of the kinetics of recombination at the ade2 locus in rad50 cells would be necessary to test this explanation. It is also possible that the contrasting effect of the rad50 mutations on radiation sensitivity and HO-induced recombination is a function of the number of DSBs produced in each case. As the number of lesions increases from two (in our recombination assay) to several (at high X-ray doses), the repair capacity of rad50 cells may become saturated. The meiotic and mitotic properties of X R S 2 and RAD50 are very similar, suggesting that both genes are similarly involved in DNA repair and recombination (Ivanov et al. 1994). This similarity extends to our recombination assay, since a strain containing a disrupted xrs2 allele,

537 lO( 1(

o~

-

-~

0,1

"~

0.01

,

0.001

• o

YME2 (wt) mutant 15

•

mutant 36

~,

mutant srs2-41

[]

YME6 (rad52::LEU2)

0.0001 10

0

10

2o

30

40

50

krads 100

10 "*

"?'

1

E

°'i 0

b

"l

20

1

40

60 80 J m -2

•

YME2 (wt)

o

mutant 15

• & --

mutant 36 mutant srs2-41 YME7 (radl::URA3)

100 120

Fig. la, b Survival following irradiation with X-rays (a) and UV (b) of mutants 15, 36 and srs2-41. The survival curves of YME2, YME6 and YME7 are included for comparison. Cells were plated and immediately exposed to the indicated radiation doses; after 2 days at 30° C, the number of colonies in irradiated and unirradiated control plates was determined

like the tad50 strain, also caused about a four-fold reduction in gene conversion to ADE2 (data not shown). At the X-ray dose used in Table 5, the sensitivity of mutant 15, mutant 36 and srs2-41 to X-rays differed only marginally from wild-type. The X-ray and the UV repair defects became more apparent at higher doses (Fig. 1). At 40 krads, mutants 15, 36 and srs2-41 were 5-10 times more sensitive than the parental strain YME2. At this dose, YME5, a strain containing a rad52 disruption allele, was four orders of magnitude more sensitive (Fig. la). At the highest UV dose tested (100 J/m2), mutants 15 and 36 were only about two-fold more sensitive than the parental strain, however, the srs2-4l mutant was approximately 10-fold more sensitive than YME2 (Fig. lb) but still three orders of magnitude more resistant than YME7 (a strain carrying a radl disruption allele). The X-ray and UV survival curves presented for the srs2-41 strain are similar to those obtained for an srs2 deletion allele (data not shown) and for some previously identified srs2 alleles (Aboussekhra et al. 1989; Palladino and Klein 1992).

The weakest alleles isolated caused a reduction of about four- to five-fold in the number of ADE-- recombinants. This indicates that the sensitivity of our assay is high, allowing the detection of mutations causing only small reductions in the ability to recombine. At least in the case of the srs2-41 allele, the increased sensitivity provided by the use of YME2 instead of YME3 as the parental strain for our screen proved to be advantageous. Thus, in YME2, which contains the HO site at both the M A T locus and the plasmid, the defect caused by a disruption of the SRS2 gene (which caused the same 10-fold reduction as the srs2-41 allele) was readily detectable. But in YME3, which has only a single HO site because it lacks the M A T locus, the effect of the srs2 disruption allele was more difficult to detect visually (data not shown). Thus, the srs2-41 allele might have been missed had the mutant screen used YME3. Also, mild alleles in rad genes like the rad52-30, rad52-35 and rad55-30 alleles might not have been detected in the YME3 genetic background. In summary, we set out to identify functions involved in HO-induced plasmid-chromosome recombination in yeast and recovered a collection of new alleles in previously identified genes and a novel rfal allele (Firmenich et al. 1995); a group of nine mutations comprising at least two comptementation groups remain to be characterized. Some of the genes found in our screen (for example, RAD52, RAD54, RAD55, RAD57) have not been assigned a clear role in recombination yet, despite the fact that they have been known for some time. We believe that our collection of alleles in each of these genes may provide new insight into their role in recombination and repair. Having described the properties of the mutants we have isolated and announced their availability, we expect that they will be useful to the recombination community. Acknowledgements We are grateful to John McCusker for providing strains and for helpful discussions and to Marianne Dieckmann who participated in the early studies leading to this work. We are also grateful to Christine Lohrlein, James Haber, Errol Friedberg, Rodney Rothstein, David Schild, Robert Schiestl and Wolfram Siede for the gifts of strains and/or plasmids. This work was supported by grant AGO-2908-12 from the National Institute of Health. M.E.-A. was a postdoctoral fellow supported by the Spanish Ministry of Education and Science. A.A.F. was the recipient of a Howard Hughes Medical Institute predoctoral fellowship.

References Aboussekhra A, Chanet R, Zgaga Z, Cassier-Chauvat C, Heude M, Fabre F (1989) RADH, a gene ofSaccharomyces cerevisiae encoding a putative DNA helicase involved in DNA repair. Characteristics of radH mutants and sequence of the gene. Nucleic Acids Res t7:7211 7219 Aboussekhra A, Chanet R, Adjiri A, Fabre F (1992) Semidominant suppressors of Srs2 helicase mutations of Saccharomyces cerevisiae map in the RAD5I gene, whose sequence predicts a protein with similarities to procaryotic RecA proteins. Mol Cell Biot 12:3224-3234

538 Basile G, Aker M, Mortimer RK (1992) Nucleotide sequence and transcriptional regulation of the yeast recombinational repair gene RAD51. Mol Celt Biol t2:3235-3246 Becker DM, Guarente L (199t) High-efficiency transformation of yeast by electroporation. Methods Enzymol 194:182-187 Bennett CB, Lewis AL, Baldwin KK, Resnick MA (1993) Lethality induced by a single site-specificdouble-strand break in a dispensable yeast plasmid. Proc Natl Acad Sci USA 90:5613-5617 Brill SJ, Stillman B (1991) Replication factor-A from Saccharomyces cerevisiae is encoded by three essential genes coordinately expressed at S phase. Genes Dev 5:1589-1600 Dorfman B (1969) The isolation of adenylosuccinate synthetase mutants in yeast by selection for constitutive behavior in pigmented strains. Genetics 61 : 377-389 Esposito MS, Brown JT (1990) Conditional hyporecombination mutants of three REC genes of S. cerevisiae. Curr Genet 17 : 7 12 Esposito MS, Hosoda J, Golin J, Moise H, Bjornstad K, Maleas D (1984) Recombination in Saccharomyces cerevisiae: REC gene mutants and DNA-binding proteins. Cold Spring Harbor Symp Quant Biol 49 : 41-48 Firmenich AA, Redding K (1993) An efficient procedure for multiple transformations of yeast in parallel. Biotechniques 14: 712-4, 716-8 Firmenich AA, Elias-Arnanz M, Berg P (1995) A novel allele of Saccharomyces cerevisiae RFA1 deficient in recombination and repair, and suppressible by RAD52. Mol Cell Biol 15:1620-1631 Fishman-Lobell J, Haber J (1992) Removal of nonhomologous DNA ends in double-strand break recombination: the role of the yeast ultraviolet repair gene RADI. Science 258:480-484 Hays SL, Firmenich AA, Berg P (1995) Complex formation in yeast double-strand break repair: participation of Rad51, Rad52, Rad55, and Rad57 proteins. Proc Natl Acad Sci USA 92: 6925-6929 Heyer WD, Rao MR, Erdile LF, Kelly TJ, Kolodner RD (1990) An essential Saccharomyces cerevisiae single-stranded DNA binding protein is homologous to the large subunit of human RP-A. EMBO J 9 : 2321-2329 Hoekstra MF, Liskay RM, On AC, De Maggio A J; Burbee DG, Heffron F (1991) HRR25, a putative protein kinase from budding yeast: association with repair of damaged DNA. Science 253: 1031-1034 Ivanov EL, Sugawara N, White CI, Fabre F, Haber JE (1994) Mutations in XRS2 and RAD50 delay but do not prevent mating-type switching in Saccharomyces cerevisiae. Mol Cell Biol 14: 3414-3425 Jensen RE, Herskowitz I (1984) Directionality and regulation of cassette substitution in yeast. Cold Spring Harbor Symp Quant Biol 49 : 97-104 Jones EW, Magasanik B (1967) Phosphoribosyl-5-amino-4imidazolecarboxamide formyltransferase activity in the adeninehistidine auxotroph ad3 of Saccharomyces cerevisiae. Biochem Biophys Res Comm 29:600-604 Kowalczykowski SC, Dixon DA, Eggleston AK, Lauder SD, Rehrauer WM (1994) Biochemistry of homologous recombination in Escherichia coil. Microbiol Rev 58:401-465 Malone RE, Esposito RE (1980) The RAD52 gene is required for homothallic interconversion of mating types and spontaneous mitotic recombination in yeast. Proc Natl Acad Sci USA 77: 503-507 Malone RE, Ward T, Lin S, Waring J (1990) The RADSO gene, a member of the double-strand break epistasis group, is not

required for spontaneous mitotic recombination in yeast. Curr Genet 18:t11-116 Mazlen AS (1968) A biochemical study of the adenine and histidine requiring mutant adenine-3 of Saccharomyces cerevisiae. Ph.D. thesis, Brooklyn College, The City University of New York Ogawa T, Yu X, Shinohara A, Egelman EH (1993) Similarity of the yeast RAD51 filament to the bacterial RecA filament. Science 259:1896-1898 Palladino F, Klein H (1992) Analysis of mitotic and meiotic defects in Saccharomyces cerevisiae SRS2 DNA helicase mutants. Genetics 132 : 23-37 Petes TD, Malone RE, Symington LS (1991) Recombination in yeast. In: Jones EW, Pringle JR, Broach JR (eds) The molecular biology of the yeast Saccharomyces cerevisiae. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York pp 407-522 Resnick MA, Martin P (1976) The repair of double-strand breaks in the nuclear DNA of Saccharomyces cerevisiae and its genetic control. Mol Gen Genet 143 : 119-129 Rodarte-Ramon U, Mortimer RK (1972) Radiation-induced recombination in Saccharornyces: isolation and genetic study of recombination-deficient mutants. Radiat Res 49 : 133-147 Rong L, Palladino F, Aguilera A, Klein HL (1991) The hyper-gene conversion hpr5-1 mutation of Saccharomyces cerevisiae is an allele of the SRS2/RADH gene. Genetics 127:75-85 Rose MD, Novick P, Thomas JH, Botstein D, Fink GR (1987) A S. cerevisiae genomic plasmid bank based on a centromere-containing shuttle vector. Gene 60:237-243 Rudin N, Haber JE (1988) Efficient repair of HO-induced chromosomal breaks in Saccharomyces cerevisiae by recombination between flanking homologous sequences. Mol Cell Biol 8: 3918-3928 Scherer S, Davis RW (1979) Replacement of chromosome segments with altered DNA sequences constructed in vitro. Proc Natl Acad Sci USA 76:4951-4955 Schiestl RH, Gietz RD (1989) High-efficiency transformation of intact yeast cells using single stranded nucleic acids as a carrier. Curr Genet 16 : 33%346 Schiestl RH, Prakash S, Prakash L (1990) The SRS2 suppressor of tad6 mutations of Saccharomyces cerevisiae acts by channeling DNA lesions into the RAD52 DNA repair pathway. Genetics 124:812831 Sherman F, Fink GR, Hicks JB (1986) Methods in yeast genetics: a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York Shinohara A, Ogawa H, Ogawa T (1992) Rad51 protein involved in repair and recombination in S. cerevisiae is a RecA-like protein. Cell 69 : 457-470 Story RM, Bishop DK, Kleckner N, Steitz TA (1993) Structural relationship of bacterial RecA proteins to recombination protein from bacteriophage T4 and yeast. Science 259:1892-1895 Sugawara N, Ivanov EL, Fishman-Lobell J, Ray BL, Wu X, Haber JE (1995) DNA structure-dependent requirements for yeast RAD genes in gene conversion. Nature 373:84-86 Szostak JW, Orr-Weaver TL, Rothstein RJ, Stahl FW (1983) The double strand break model for recombination. Cell 33 : 25-35 Weiffenbach B, Haber JE (1981) Homothallic mating type switching generates lethal chromosome breaks in rad52 strains of Saccharomyces cerevisiae. Mol Cell Biol 1 : 522-534 Weinert TA, Hartwell LH (1988) The RAD9 gene controls the cell cycle response to DNA damage in Saccharomyces cerevisiae. Science 241 : 317-322