Genetic Control of RNA Polymerase I-Stimulated Recombination in Yeast

Betty R. Zehfus,* Andrew D. McWilliams," Yu-Huei Lin,* Mer1 F. Hoekstrat and Ralph L. Keil* *Department of Biological Chemistry and Intercollege Program in Genetics, The Milton S. Hershey Medical Center, Hershey, Pennsylvania 17033, and 'Molecular Biology and Virology Laboratories, The Salk Institute for Biological Studies, La Jolla, Calqornia 920?7 Manuscript received November 6, 1989 Accepted for publication May 12, 1990

ABSTRACT We examined the genetic control of the activity of HOTl, a cis-acting recombination-stimulatory sequence of Saccharomyces cerevisiae. Mutations in RADl and RAD52 decrease the ability of HOTI to stimulate intrachromosomal recombinationwhile mutations in RAD4 and RAD50 do notaffect HOTl activity. In radlA strains, the stimulation of excisive recombination by HOTl is decreased while the rate of gene replacement is not affected. In rad52-8 strains the ability of HOTl to stimulate both excisive recombination and gene replacement is decreased. All of the recombinants in the rad52-8 strainsthat would be categorized as genereplacements based ontheir phenotype are diploids apparently derivedby endomitosis and excisive recombination. Studies on radlA rad52-8 strainsshow that these mutations interact synergistically in the presence or absence of HOTI, resulting in low levels of recombination. The rate of gene replacement but not excisive recombination is stimulated by HOTl in radlA rad52-8 strains. Taken together, the results show that HOTl stimulates exchange using multiple recombination pathways. Some of the activity of HOTl is RADI-dependent, some is RAD52-dependent, and some requireseither RADl or RAD52 assuggested by the synergistic interaction found in double mutant strains. There is also a component of HOTl activity that is independent of bothRAD1 and RAD52.

HE rate of homologousrecombination is not Transcription and recombination are also thought T uniform throughout the genome of an orga- to be associated for some site-specific and meiotic nism. One factor that affects the rate of mitotic ex- recombination events. In yeast, only transcriptionally change is transcription. KEILand ROEDER(1 984)iden- active copies of the three mating-type loci are cleaved tified a cis-acting recombination hotspot, HOTl, that by the HO endonuclease (KLAR, STRATHERNand stimulates mitotic recombination when inserted at ABRAHAM1984), permitting them to function as re- novel locations in the yeast genome. This hotspot is cipients during the switching event. In mammalian part of the ribosomal DNA (rDNA) repeat unit of cells rearrangement of immunoglobulin and T cell yeast. The fragmentthat stimulates recombination receptor genes is enhanced by transcription (BLACK- contains theenhancer (ELIONand WARNER1984, WELL et al. 1986; YANCOPOULOSet al. 1986; SCHLISSEL 1986) and transcription initiation site (KLEMENZand and BALTIMORE1989). BLACKWELLet al. (1986) pos- GEIDUSCHEK1980; BAYEVet al. 1980)for the 35s tulated that transcription may alter theaccessibility of rRNA. Recombination is stimulated only when the these genes for rearrangement. NICOLASet al. (1 989) fragment is oriented such that RNA polymerase I showed that the promoter region of the yeast ARG4 transcription initiated in the fragment could proceed gene contains an initiation site for meiotic gene con- across the recombining sequences. Subcloning (VOELKEL-MEIMAN,KEIL and ROEDER 1987)and version. Double-strand breaks occur in this region at linker insertion (STEWARTand ROEDER1989) studies the time of recombination as well as near two other showed that approximately the same regions are re- promoters that were studied (SUNet al. 1989). quiredboth for highly efficient, properlyinitiated Numerous mutations in yeast affect DNA recom- transcription and for recombination hotspot activity. bination and repair. Analysis of strains containingtwo In addition, insertionof the transcription termination or more of thesemutations defines three epistasis site (KEMPERS-VEENSTRAet al. 1986) between HOTl groups based on sensitivity to ultraviolet (UV) or and the recombining sequences abolishes the recom- ionizing radiation orto mutagenic chemicals (re- bination-stimulatory activity of HOTI (VOELKEL-MEI- viewed by HAYNFS and KUNZ 1981). The epistasis MAN, KEIL and ROEDER1987). THOMASand ROTH- groups have been termed (1) error-freeexcision repair STEIN (1989a) found that RNA polymerase I1 tran- of pyrimidine dimers or the RAD3 group, (2) double- scription also stimulates mitotic recombination strand-breakrepair orthe RAD52 group,and (3) between directly repeated sequences. error-prone repair or the RAD6 group. We studied

Genetics 126 41-52 (September, 1990) 42 B. R. Zehfus et al. the effects of mutations in genes from the RAD? and termed E1 + HOTI) wereused to create recombination RAD52 groups on HOTlactivity. substrates (Figure 1, A and B, respectively) as described previously (VOELKEL-MEIMAN,KEIL and ROEDER1987). MONTELONE, HOEKSTRAand MALONE (1988) found To further characterize the effects of radl and rad52 that mutations in either RADl or RAD4, members of mutations on HOTl-stimulated recombination, the strain the RAD? group, increase the frequency of interchro- RLK88-3C (MATa leu2-3,I 12 his#-260 ura3-52 ade2-1 mosomal crossing overbut not gene conversion. They trpl-HIII lys2-ABX canr) was used. The trpl-HIII and lys2- also found thata mutation in RADl or RAD4 abolishes ABX mutations have been described previously (YUANand KEIL 1990). Plasmids lacking HOT1 (pL524), containing E1 the enhanced mitotic gene conversion phenotype of + HOTl (pL623), containing HOTI on a 4.6-kb BglII-B reml, an allele of RAD?, but not the increased level fragment of rDNA, termed BglII-B HOTI (pL559), and of crossing over in rem1 strains. Mutations in RADl containing the inactive orientation of the 570-bp subclone reduce the rate of certain intrachromosomal recom- of HOTl (pL625, called IE c HOTI) were introduced by bination events during mitosis (SCHIESTLand PRAK- transformation into this strain to create recombination substrates. Three independent transformants containing each ASH 1988;KLEIN 1988; THOMASand ROTHSTEIN plasmid properly integrated (Figure 1, A, B, C, and D, 1989b). The preciseroles of the RADl and RAD4 respectively) were identified (VOELKEL-MEIMAN,KEIL and products in recombination is unclear but it has been ROEDER1987). A deletion-disruption mutation of the RADl suggested that they are involved in forming or proc- gene, radlA::LEU2 (called radl A), was constructed by de- essing certain intermediates for recombination-repair leting a 2.1-kb StuI-ClaI-ClaI fragment of RADI, filling the 3' recessed end by the activity of the Klenow fragment of (MONTELONE,HOEKSTRA and MALONE 1988; Escherichia coli DNA polymerase I and inserting the LEU2 SCHIESTLand PRAKASH1988; KLEIN 1988). gene on a 3.2-kb BglII fragment that was made blunt ended Genes of the double-strand-break repair group are with Klenow fragment. The one-step gene disruption tech- alsoinvolved in mitotic recombination. Numerous nique (ROTHSTEIN 1983) was used to insert the radlA mu- reports document that rad52 mutations reduce the tation into appropriate strains. The rad52-8::TRPI (called rad52-8) mutation (obtained from D. SCHILD)was inserted rate of several forms of mitotic recombination. These into strains by the same procedure. Southern analysis and includeinterchromosomal (PRAKASH et al. 1980; phenotypic characterization were used to confirm that GAMEet al. 1980; MALONEand ESPOSITO 1980) and proper disruptions had occurred. intrachromosomal (JACKSON and FINK 198 1) mitotic The E. cola strain MC1066 [led trpc pyrF::TnS (KanK) recombination.In addition, RAD52 is required for araT lacX74 del strA hsdR hsdM) obtained from M. CASA- DABAN was used throughout this work. mating-type switching (MALONEand ESPOSITO1980), DNA manipulations: Restriction endonucleases and andrepair of double-strand breaks (RESNICKand DNA modification enzymes were purchased from several MARTIN 1976).MALONE and ESPOSITO(1981) and sources and used according to the manufacturer's specifi- MALONE (1983) found that mutations in the RAD50 cations. DNA manipulations were performed as described gene, another memberof the RAD52 group, increase previously (KEILand ROEDER1984). Media and growth conditions: Yeast synthetic complete the frequency of mitotic interchromosomal recombi- (SC) and sporulation media were as described by MALONE, nation.The mechanism by which RNA polymerase I GOLINand ESPOSITO(1 980) except that for SC medium L- transcription stimulates recombination is not known. aspartic acid (74 mg/liter) and valine (140 mg/liter) were To determine whether HOTl-stimulated mitotic re- added and the pH was not adjusted. Drop-out media (e.g., SC-his) were prepared by omitting the appropriate constit- combination requires at least some of the same gene uent from SC medium. To select Ura- cells, 0.8 g of 5- products as other mitotic exchange, we have studied fluoro-orotic acid (5FOA) per liter was added to SC medium. the effects on HOTlactivity of several rad mutations. The 5FOA was either synthesized (ALAM,SHIRES and The results indicate that a number of exchange path- ABOUL-ENEIN1975) or purchased from PCR Incorporated. ways are involvedin producing the recombination- Other yeast media were prepared as described in SHERMAN, FINK andHICKS (1 986). Bacterial media were formulated as stimulatory activity HOTl and that atleast some of of described in MANIATIS, FRITSCHand SAMBROOK(1982). these pathways participatein exchange in the absence Strains were sporulated by growing on YPD plates overnight of HOTl. at 30°, replicating to sporulation medium and incubating at 30". MATERIALS AND METHODS Fluctuation tests, independent recombinants, and data analysis: Fluctuation testswere conducted as described Strains andplasmids: Strains containing a radl-2, rad4, previously (KEIL and ROEDER1984) except that cultures rad50-1 or rad52-I mutation are congenic isolates kindly were grown in SC medium. The number of Ura- and His+ provided by R. E. MALONE. These mutant strains were recombinants were determined by plating appropriate di- crossed to RLK1-3C (MATahis4-260 ura3-52 ade2-1 lutions of the cultures on SC + 5FOA and SC-his, respec- can'). Following sporulation, tetrads were dissected by stand- tively. The recombination frequency was determined by the ard techniques (SHERMAN,FINK and HICKS1986). From method of the median (LEAand COULSON1949). Rates were each cross a Rad- segregant containing MATa, his#-260 and determined by dividing the frequency by (In N - In NO) ura3-52 was obtained. Two further backcrosses to RLK1- where N is the final number of cells in the culture and NOis 3C were conducted for each rad mutation. From the third the initial number (DRAKE 1970).The median test as de- backcross, three Rad- and three Rad+ progeny were picked scribed by SIEGEL(1956) was used to test for significant forfurther study. Plasmids pL524 (lacking HOTI),and differences between median recombination rates. pL623 (containing a 570-bp subclone of HOT1 called EI, Independent His+ or Ura- recombinants were obtained HOTI-Stimulated Recombination 43 by replicating SC plates containing approximately 100 col- absence of HOTl. Rad4 does not affect intrachromo- onies to SC-his and SC + 5FOA. Recombinants from each soma1 recombination in the presence or absence of colony were streaked on SC and a single colony from this streak was tested for appropriate phenotypes. To test for HOTl. significant differences between the proportion of different Mutations in both RAD50 and RAD52, members of types of recombinants produced by the various genotypes, the double-strand-breakrepair group, were also the 2 x 2 contingency test was used (SNEDECORand COCH- tested. Rad52-1 decreases the rate of both His+ and RAN 1967). Ura-recombination in the presence or absence of HOTl. The rad50-1 mutation does not affect recom- RESULTS bination in the presence or absence of HOTl. The strains used in the above studies were derived Assay for the recombination-stimulatory activity by three backcrosses. However, there is still consid- of HOTl: Wheninserted at novel locations in the erable variation in the level of recombination in the yeast genome, HOTlstimulates mitotic recombination strains. For example, there is about a sevenfold differ- in adjacent sequences (KEIL and ROEDER1984). Two ence in the rate of His+ recombination in the RAD50 versions of HOTl were used in theseexperiments. and RAD52 strains containingE1 +HOTl. To further One is the 4.6-kbp BgZII-B fragment of rDNA, termed study the effects of radl and rad52 on the recombi- BglII-B HOTl, that contains the 3' half of the 25s nation-stimulatory activity of HOTl we constructed rRNA gene, thenontranscribed spacer, the 5s rRNA isogenic strains containing these mutations by yeast gene, and the 5' sequences encoding the 35s pre- transformation. rRNA. The other version, E1 + HOTl(or IE c The data in Table 2 show that the deletion-disrup- HOTl when inserted in the inactive orientation), is a tion mutation radlA significantly decreases the rate 570-bp subclone derived from the BgZII-B fragment of HOTl-stimulated recombination in almost all cases. that produces a larger stimulation of recombination This decrease is approximately two- to fivefold for than the BgZII-B fragment. This version contains the both His+ and Ura- recombination. In strains without enhancer and initiation site for the 35s rRNA tran- HOTl at his4 or strains containing the inactive orien- script, and the arrow indicates the direction of RNA tation, IE t HOTl, the radlA mutation doesnot polymerase I transcription. These HOT1-containing affect the rate of recombination. In strains containing fragments are inserted into a plasmid thatcan be the rad52-8 mutation there is a significant decrease in targeted to integrate at thehis4 locus on chromosome the rate of both His+ and Ura- recombination in the ZZZ. Integration of these plasmids produces a duplication of his4 alleles separated by pBR322 and URA3 presence or absence of HOTl. sequences (Figure 1). Recombinants that are His+ or Mutations in RADl and RAD 52interact synergis- Ura- can be produced from these duplications. Ura- tically to reduce recombination: The data from the recombinants, which can be either His+ or His- are single mutantstrains suggest thatradl specifically defined as excisive recombinants(Figure 2). These affects HOTl-stimulated recombination while rad52 can result from a numberof different events including has a general effect on mitotic intrachromosomal re- simple intrachromatid reciprocal exchange, intrach- combination at HZS4. To examine the interaction of romatid gene conversion associated with a reciprocal thesemutations, we constructedstrains containing cross-over, unequal-sister-chromatid exchange, or un- both radlA and rad52-8. As shown in Table 2, the equal-sister-chromatid conversion. His+Ura+recom- rate of both His+ and Ura- recombination decreases binants are produced by events termed gene replace- significantly in the double mutant strains as compared ments. These recombinants contain bothcopies of the to the single mutant strains. The only case in which repeated chromosome ZZZ sequences, and the plasmid this does not occur is the production of Ura- cells in and URA3 sequences. Gene replacements can be pro- the absence of HOTl. The synergistic interaction of duced by events including intrachromatid or sister- radl with rad52 in the presence or absence of HOTl, chromatid gene conversion unassociated with recip- indicates that the effect of radl on recombination is rocal exchange or double reciprocal exchange. not specific to HOTlactivity. Mutations in RADl and RAD52 decrease the re- Further characterizationof recombinants: To fur- combination-stimulatoryactivity of HOTl: Muta- ther examinerecombination in thesestrains, inde- tions in two differentgenes involved in error-free pendent recombinants were isolated from strains that excision repair (reviewed by HAYNESand KUNZ198 1) lack HOTl andstrains containing E1 +HOTl. Differ- were tested for their effect on HOTl-stimulated re- ent types of events can lead to the productionof His+ combination. As shown in Table 1, the radl-2 muta- recombinants.Recombination between the repeats tion significantly decreases the recombination-stimu- can lead to the production of His+Ura+ genereplace- latory activity of E1 + HOTl. However, as shown by ments and Ura- excisive recombinants that are either the results for strains lacking HOTl, radl-2 does not His+ or His- (Figure 2). Other events can also lead to significantly alter the rate of recombination in the the production of cells with these phenotypes. South- 44 B. R. Zehfus et al. FIGUREI .-Recombination substrates. Duplications of two different his4 nluta- tions separated by pBR322 and URA3 sequences were constructed as described previously (VOELKEL-MEIMAN,KEIL and ROEDER 1987). One of the duplications (A) lacks HOTI. The duplication in (B) contains a 570-bp subclone of HOTl ori- B hid-A UR43 hid-260 ented (El+) such that it will stimulate A. noHOTl k\\N\\\Y I* I exchange between therepeated his4 sequences. The duplication in (C) contains + E1 hi&A URAS hid260 HOTI in the 4.6-kbp BglIl-B fragment of rDNA. The duplication in (D) contains the B. EI4fOTl I I-h\\\\\\V-l I* 570-bp subclone of HOTI in the inactive orientation (IEt) that does not stimulate -c Bgm-B hi&A hid260 recombination of the repeated sequences. C. BgnI-B HOTl I* Heavyline: chromosome 111 sequences near his4; open block: sequences 5’ of his4 that are duplicated; dotted block: his4 + IE hid-A URAS hid-260 gene with the arrowhead representing the D. IEcHOTl I* 3’ end: thin line: pBR322 sequences; di- agonally striped block: URA3 gene; filled block: subclones of rDNA that contain HOT]; star: indicates the relative position of the his4260 mutation; the arrow above the HOTI fragment indicates the active (+) or inactive (t)orientation forincreas- ing the frequency of His+and Ura- recombinants; B: the BamHl linker at which HOTl was inserted.

hi84-A URAS hid-260 I l-k\\\\\\W=l I*

FIGURE2.-Recombination events. One possible pairing arrangement is dia- grammed. Recombination producing a His+ Ura+cell that contains the duplicated his4 sequences flanking the plasmid and URAS sequences is defined as gene replacement. Excisive recombinants are \ Ura- and have lost one of the his4 re- GENE Excram REP” RECOMBINATION peats as well as the plasmid and URA3 sequences. The symbols are described in \ the legend to Figure 1. h&4-A (IRAS HIS4 I t-K\\\\\\W I d m+um+ ~is’um-or ern analysis was performed to determine whether the his4 alleles separated by plasmid and URA3 sequences, chromosomal configuration of independent His+Ura+ while the other copy of chromosome ZZZ has under- and Ura- recombinants was consistent with gene re- gone excisive recombination. To determine whether placement and excisive events, respectively. We define these recombinants were disomic (n + 1) or MATa/ gene replacements as His+Ura+cells that contain the MATa diploids, they were crossed to a haploid and a repeated chromosome ZZZ sequences and the plasmid MATaIMATa diploid. Four His+Ura+ recombinants and URA3 sequences on a single chromosome (Figure from the rad52-8 strains lacking HOTl and threesuch 2). The His+Ura+ recombinants from the wild-type recombinants from the rad52-8 strains containing and radlA strains appear to have resulted from gene HOTl were tested. All seven recombinants gave fewer replacement (Table 3). However, regardless of the than 10% viable spores when crossed to ahaploid but presence or absenceof HOTl, Southern analysis more than 70% viable spores whencrossed to a showed that all 20 His+Ura+recombinants examined MATaIMATa diploid. Furthermore, in the tetraploids from rad52-8 strains contain two copies of chromo- resulting from the latter cross, markers such asADEZ some ZZZ (Table 3). One of the copies of chromosome segregated 2+:2-, 3+:1-, and 4+:0- as expected from ZZZ contains the parental configuration of the repeated the genotype of the tetraploid (ADE2/ADE2/ade2/ HOTI-Stimulated Recombination 45

TABLE 1

Effect of repair and recombination mutantson the rate of HOTI-stimulated recombination

No HOT1 rate (XIO') El + HOTl rate (X105)

Strain His+ Ura- His+ Strain His+ urd-

rad 1-2 1.07 (9) 0.60 (9) 4.63 (18)** 22.4(9)** RAD I 1.2 1 (9) 0.74 (9) 53.6(18) 196 (9) rad4 1.34 (9) 0.88 (9) 53.1 (9) 254 (9) RAD4 (9) 1.38 37.5 (9) rad50- 1 0.98 (15) 0.53 (1 5) 178 (17) 42.4 (17) HA1150 1.69 (18) 35.0 (18) rad52-1 (9)** 0.69 0.41 (9)** 0.26 (9)** 2.74 (9)** RAD52 1.26 3.80(9) (9) 235 (9) 238 (9) The rate of Ura- recombination for rad4 and rad50-1 was compared to be combined median rate of Ura- recombination for the 18 RAD1 and RAD52 cultures tested. For all other comparisons the recombination rates for the mutant strain were compared to the rates for the appropriate wild-type strain. The number of cultures tested for each strain is shown in parentheses. * Significantly different at P = 0.05; ** significantly different at P = 0.01.

TABLE 2

Effect of radlA and rad52-8 on the rateof HOTI-stimulated recombination

BglII-B HOTl rate IE c HOTl rate No HOTl rate (X IO') E1 + HOTl rate (XIO?) (X 1 OS) (X 105)

Strain His+ Ura- His+ Ura- His+ U ra- His+ u rd- Rad+ 2.00 1.63 6.27 7.72 90.1 360 1.48 2.81 radlA 1.42 1.64 2.68** 4.80 18.5** 120** 1.12 2.19 rad52-8 0.30** 0.65* 0.21** 1.82** 0.35** 3.97** 0.10** 1.03* radlA, rad52-8 0.017** 0.26 0.023** 0.18** 0.10* 0.55** 0.01* 0.13* The rate of recombination was determined by assaying 27 cultures for the wild-type strains, 18 cultures for the radlA and rad52-8 strains and 9 cultures forthe rudIA rad52-8strains. Statistical significance of the data for theradlA and rad52-8 strains is indicated for comparisons of the recombination rates for these mutant strains to the wild-type strains. Significance for comparisons of the data for the radlA rad52-8 strains to the rad52-8 strains is shown. The symbols are as described for Table 1. ade2). Thus, these cells are diploids apparently de- diploid His+Ura+recombinants is an allele-specific rived by endomitosis and excisive recombination and phenotype of the rad52-8 disruptionmutation, will be called His+Ura+ diploids. In crosses involving His+Ura+ recombinants derived fromrad52-1 strains these His+Ura+ diploids, the His+ and Ura+ pheno- were tested.Southern analysis showed that two of types oftensegregated todifferent spores showing four such recombinants from a strain lacking HOTl that the HIS4 and URA3 genes are in repulsion. and one of fourrecombinants from a strain containing In the radlA rad52-8 strains only a few (4 of 20) of E1 +HOTl contained two copies of chromosome ZZZ. the His+Ura+ recombinants were diploid. In the ab- The rad52-8 and the rad52-1 strains used were not sence of HOTl, 3 of 12 His+Ura+ recombinants were isogenic. Therefore, this difference may be due to a diploids, while one of eight such recombinants from difference between the rad52-8 and rad52-I alleles the strains with HOTl was a diploid (Table 3). Genetic or to other genetic differences between the strains analysis of two of these recombinants showed that the used. HIS4 and URA3 genes were in repulsion. The haploid His+ excisive recombinants are Ura- and should His+Ura+ recombinants in the radlA rad52-8 strains have lost one copy of the his4 repeat as well as the do not appear to result from spontaneous mutation. plasmid and URA3 sequences. Southern analysis In radlA rad52-8 strains therate of spontaneous showed this had occurred forall of thestrains. Tetrads reversion of his4-260 to His+ (0.8 X lo-' in the derived from crossing His+Ura- recombinants from absence of HOTI and 2.8 X IO-' when E1 +-HOTI is the rad52-8 and radlA rad52-8 strains to a haploid present) is more than ninefold lower than the rate of gave greater than 80% spore viability, and the chro- His+Ura+recombination (see below andTable 5). mosome ZZZ markers segregated 2:2 showing that the Thus most of the haploid His+Ura+cells in the radlA recombinants were not diploid or disomic. rad52-8 strains result from recombinationrather than Ura- recombinants can be either His+ or His-. In mutation. both cases excisive recombinants are expected to lose To investigate whetherthe production of these one of the his4 repeats and the plasmid and URA3 46 B. R. Zehfus et al.

TABLE 3 Summary of molecular and genetic characterization of events

Number of independent recombinants His+Urd+ Ura-His-

Gene Ura-His+ Strain replacements Diploids Excisive Excisive Mutations No HOT1 Kad+ 17 0 8 4 0 radlA 12 0 7 3 0 rad52-8 0 14 8 4 0 radlA rad52-8 9 3 7 6 8 E1 + HOTI kdd+ 8 0 6 4 0 radlA 8 0 8 3 0 rad52-8 0 6 8 4 0 radlA rad52-8 7 1 5 6 7 Independent His' and Ura- colonies were characterized bv Southern analysis and genetic techniques as described in the text to classify the event that produced the putative recombinant. sequences. All of the independent Ura- recombinants an increase in the proportion of independent His+ from wild-type, radl A,and rad52-8 strains as well as recombinants that are Ura+. Thus rad the 1 A mutation the Ura-His+ recombinants from the radlA rad52-8 decreases the proportion of His+ recombinants that strains showed the pattern expected for excisive re- are formed by excisive recombination in the presence combinants.From the radlArad52-8 strains, how- or absence of HOTl. This finding further shows that ever, about55% of the independent Ura-His- recom- rad 1 A affects recombination in the presence or ab- binants still contained the parental configuration of sence of HOTl. the duplication at his4. Two lines of evidence indicate The results from the rad52-8 strains generally show that these cells result from spontaneous mutations in a decrease in the proportion of His+ recombinants the LIRA3 gene on chromosome III. First, Southern that are Ura+ as compared to wild type. This is not blot analysis shows that the restriction maps of the true for the strains containing E1 + HOTl, which recombinant and parental chromosomesare identical, have avery low proportion of His+Ura+ recombinants indicating they were not derived from ectopic recom- even in the wild-type strains. However, as described bination with the ura3-52 mutation on chromosome above,these His+Ura+ recombinants from rad52-8 V, which contains a Ty element (ROSEand WINSTON strains are not derived by gene conversion. As shown 1984). These results also show that these Ura- cells by the numbers in parentheses, all of the His+Ura+ result from point mutations or other small mutations. recombinants tested in rad52-8 strains containing E1 Second, these strains fail to complement a ura3 strain -+ HOTl or lacking HOTl are excisive recombinant but they do complement a strain containing a ural diploids. We assume this is also true for the rad52-8 mutation. As expected, these Ura-His- recombinants strains containing BglII-B HOTl or HOTl. still give rise to His+ recombinants at the rateobserved In the radlA rad52-8strains the proportion of His+ for rad I A rad52-8 strains. recombinants that are Ura+ is approximately 50% in The proportions of different types of recombi- the presence or absence of HOTl. As shown in Table nants are altered by HOTl, radl and rad52: Inde- 4A, this proportion is in general different from that pendent His+ recombinants were isolated and tested found in strainscontaining either single mutation. to determine if they were Ura+ or Ura-. The results The proportion of these recombinants that are dip- from this analysis are shown in Table 4A. For the loids rather than gene replacements is shown in pa- wild-type strains there is no difference in the propor- rentheses. tion of His+Ura+recombinants in the absence of Ura- excisive recombinants can be either His+ or HOTl and when BgZII-B HOTl is present. However, His-. As shown by the datain Table 4B the proportion when E1 + HOTl or IE t HOTl is present there is a of the Ura- recombinants thatare His- in Rad+ strains significant decrease in theproportion of His+Ura+ is altered by the presence of HOTl in the active recombinants. Thus, when IE t HOTl is present it orientation either in the BgZII-B fragment or in E1 + does not affect the rate of His+ recombination (Table HOTl. In therad 1 A strains this proportion is changed 2) but it does significantly alter the proportionsof the only in the presence of BgZII-B HOTl. In the rad52-8 different events. strains the proportion of Ura- recombinants that are As compared to wild type, the radlA strains show His- is higher than the proportion in wild-type strains H O TI-Stimulated Recombination HOTI-Stimulated 47

TABLE 4

Effect of radlA and rad52-8 on typesthe of recombinants formed

radlA rad524 radlA rad524 radlA rad524Rad+ radlA

A. Percent of independent His' recombinants that are Ura' No HOTI 54.2 75.3** 24.1** (100%) 51.5**, ** (25%) BgllI-B HOTI 48.6 94.7** 16.1** (ND) 41.3**. ** (ND) (1 El + HOT1 10.0** 44.7** 14.1 (100%) 55.4~51).*L 2.5%) 1 E c HOT I 31.9** 76.0** 14.3** (ND) 58.9*, ** (ND) B. Percent of' independent Urd- recombinants that are His- No HOTl 78.7 85.6 88.0 92.7" BglII-B HOTI 60.7** 90.0** 97.2** 89.5" El + HOT1 88.3* 92.5 97.7** 96.0" IE + HOT1 87.9 94.0 96.2* 94.1" More than 50 independent reconlbinants were scored for each genotype. Statistical significance for the various Radf strains is indicated f~)rcotnparisons of the HOTI-containing strains to thestrain lacking HOTl. For the strains containingradlA or rad52-8 statistical significance is shown for the comparison to the appropriate wild-type strain. The data for theradZA rad52-8strains were compared to the data forradlA and rad52-X strains, respectively. The numbers in parentheses show the percent of His+Uraf recombinants that werediploid. " These numbers were corrected for the 55.6% (15 or 27) of the Ura-His- colonies that arose from spontaneous mutation as determined by So~cthernanalysis. These events occur in the presence or absence of HOTZ. Therefore, corrections were made for all of the strains even though only strains lacking HOT1 and strains containingE1 -+ HOTl were actually tested. ND = not determined; NSD = not significantly different; other symbols are as described for Table 1.

TABLE 5 Calculated rates of events

KateHis+Ura+of (X105) RateHis+Ura-of (X 10') fromRate Ura-His-of (X 1 05)

Gene Strain replacement Diploids Indep His+ Indep Ura- RecombinationMutation Ura- Indep His+ IndepDiploids replacement Strain No HOTI Rad+ 1.08 <0.06 0.92 0.35 1.28 C0.32 radlA 1.07 eo.09 0.35 0.24 1.40 <0.46 rad52-8 <0.005 0.07 0.23 0.08 0.57

~~ Rateswere calculated by multiplying the rate ofHis+ or Ura- recombination (Table 2) by the proportion of the various events as determined by molecular and genetic analysis (Tables 3 and 4). Where no events were detected, an upper limit on the rate was determined by assuming that one of the cells exanlined by Southern blot analysis (Table 3) was of the indicated type. whenever an active or inactive form of HOTl is pres- HOTI, recombinants consistent with a gene replace- ent. Strainscontaining radlA rad52-8 have a very ment event were not detected in rad52-8 strains, and high proportion of Ura- recombinants that are His- therate of excisive recombination is decreased in even after correcting for the Ura- cells produced by thesestrains compared to wild type. For excisive spontaneous mutation. events there is a synergistic interaction betweenradl A The frequency of gene replacementis not affected and rad52-8 in strains containing or lacking HOTI. in rad1 strains: The rate of various types of recom- However,this is nottrue for the production of bination can be calculated by multiplying the propor- His+Ura+ gene replacements.This is most clearly seen tions of different events from Tables 3 and 4 by the for strains containing HUTl. The rate of formation rates of His+ and Ura- recombination in Table 2. The of His+Ura+cells in the rad52-8 strains is 0.05 X results of these calculations are shown in Table 5. All six recombinants analyzed by Southern blots were Most interestingly, the rate of excisive events, both diploids, so the rate of gene replacement is less than His+Ura- and His-Ura-, is decreased by radlA in 0.01 X (therate of genereplacement if 1 of the strains containing HOTI. However, the rate of gene 6 had been derived from such an event). However, replacement events in radlA strains containing HOTl the rate of gene replacement for the double-mutant is not decreased. In strainslacking HOTl, radl A does strains is 0.05 X which is higherthan the rate not have a dramatic effect on any one category of for the rad52-8 strains. recombination event. In the presence or absence of The rate of His+Ura- recombination can be calcu- 48 B. R. Zehfus et al.

TABLE 6 Fate of HOTl or BamHI linker in recombinants

Independent recombinants His+Ura+ His+Ura- His-Ura- His+Ura- His+Ura+ Strain I copy= E copies” Multipleb +‘ - +’ - Rad+ No HOT1 10 7 0 7 1 2 2 El + HOTl 6 2 0 6 0 1 3 radlA No HOTI 7 3 2 7 0 0 3 E1 + HOTl 7 1 0 8 0 0 3 rad524 No HOTI 0 0 0 8 0 1 3 E1 + HOTl 0 0 0 8 0 1 3 radlA rad52-8 No HOT1 8 1 0 7 0 3 3 E1 + HOTl 6 1 0 5 0 1 5 Independent His’ and Ura- recombinants were characterized by Southern analysis. Diploid His+ Ura+ recombinants and spontaneous Ura- mutants were excluded from this analysis. Number of copies of BamHI linker for non-HOT1 strains or HOTl for strains containing E1 + HOTl. The BamHl linker is the site at which HOTI is inserted (Figure 1A). In all cases where there is one copy of the BamHI linker or HOTl it is in the same repeat as in the parental strain. There is at least a triplication of the his4 repeated sequences. ‘ + indicates the presence of either the BamHI linker or El + HOTl as appropriate for a given strain. - indicates the absence of these elements in the recombinants. latedfrom either the His+ data orthe Ura- data. Two of these were found in the radlA strain. Only a These two rates should be equal. However, they are relatively small number of recombinants can be read- not (Table 5). In general the rate calculated from the ily examined by these techniques. Larger numbers of Ura- data is two- to sixfold lower than the rate deter- recombinants would have to be tested to determine if mined from the His+ data. This may result from the there is a significant difference in the proportion of phenotypic lag associated with growth of ura3 cells on the types of recombinants recovered from the differ- medium containing 5FOA (RONNEand ROTHSTEIN ent genotypes. 1988; YUANand KEIL 1990). His+Ura-recombinants can bederived either as Southern analysis of independent recombinants: independent His+ or Ura- cells. Examining both cases Further informationregarding the recombination we find that almost all of them (56 of 57) contain the events can beobtained from Southern analysis of BamHI linker or HOTl. This is consistent with a DNA derived from independent recombinants. The reciprocal exchangeoccurring 3’ of the his4-260 repeated his4 genes differ by the his4-260 mutation. mutation to produce a His+ recombinant. The major- In addition, strains lacking HOTl contain a BamHI ity of independent Ura-His- recombinants lack the linker inserted at the PvuIIsite 651 bp 5’ of the HIS4 BamHI linker or HOTl. In radlA strains this is the coding sequence. This is the site at which HOTl is only class of Ura-His- recombinants observed. A re- inserted. It is possible to follow the fate of the BamHI ciprocal exchange upstream of the BamHI linker (or HOTI) would produce a Ura-His- recombinant lack- linker or HOTl in recombinants. Independent recom- ing the linker (or HOTI). The Ura-His- recombinants binants from strains lacking HOTl or containing E1 that contain the BamHI linker (or HOTl) could be HOTl were analyzed and the results are presented - produced by a reciprocal exchangethat occurred in Table 6. Examination of the His+Ura+ gene re- between the his#-260 mutation and theBamHI linker placements shows that three types of events are ob- (or HOTI ). served. The majority of these recombinants have the parental configuration of BamHI or HOTl. A second DISCUSSION type of recombinant has BamHI or HOTl present in both of the his4 repeats. We found this type of recom- We used intrachromosomal recombination between binant inall of thestrains that gave rise togene direct repeats to further characterize the recombina- replacements. The final type of gene replacement had tion-stimulatory activity of HOTI. Some, but not all, at least three copies of the repeated his4 sequence. of the genes in theerror-free excision repairand These may have been produced by a gene conversion double-strand-break repair pathways affect HOTl ac- followed by an unequal-sister-chromatid exchange. tivity. Our results show that a mutation in either rad1 HOTI-Stimulated Recombination 49 or rad52 decreases HOTl activity. HOTl stimulation reduces the ability of HOTl to stimulate plasmid ex- of plasmid excision but not gene replacement is de- cision but does not affect the increased production of creased in radlA strains. Rad52-8 decreases both gene replacementsby HOTl. THOMASand ROTHSTEIN HOT1-stimulated plasmid excision and gene replace- (1989b) saw a similar, but smaller, effect of rad I on ment. In radlA rad52-8 strains, a synergistic interac- RNA polymerase 11-stimulated recombination. In the tion leads to very low levels of recombination. HOTl absence of HOTl, radlA does not affect the rate of stimulates the formation of genereplacements in recombination but it does significantly increase the radlA rad52-8strains. These results indicatethat proportion ofHis+ recombinantsthat are genere- HOTI uses multiple recombination pathways to stim- placements (Table 4). This is also truefor strains ulate recombination. containing the inactive, IE c HOTl, orientation of Effects of HOTl on intrachromosomal recombi- HOTl. SCHIESTLand PRAKASH (1988) and KLEIN nation: In wild-type strains the E1 + HOTl subclone (1 988) found that the rateof mitotic excisive recom- stimulates both plasmid excision andgene replace- bination for nontandem duplications is decreased in ment(Table 5). However, therate of excision is radl strains. KLEIN (1988) found that the rateof gene stimulated much more (approximately 200-fold) than replacement is not affected by radl. the rate of gene replacement (about sixfold). HOTI- We find that both in the presence and absence of stimulated recombination may preferentially produce HOTl, rad52-8 abolishes gene replacement eventsand excision events. Alternatively, this finding may result also decreases the rate of excision (Table 5). Using fromthe recombination substrate used to measure duplications atthe his4 locus, JACKSON and FINK HOTl activity. The relative position of HOTl and the (1981) found that rad52 preferentially affected the his4 alleles may dramatically affect the recombination frequency of gene replacementwhile not affectingthe events that can be recovered and, therefore,influence occurrence of excisive events. KLEIN(1 988) andNICK- the level of stimulation measured. NICKOLOFFet al. OLOFF et al. (1 989) found thatexcisive recombination (1 989) studiedeffects of double-strand breaks intro- was decreased in rad52 strains. The differences be- duced by the HO endonuclease on direct-repeat re- tween these results may be due to the differentdupli- combination. Theyfound that gene replacement cations studied orto variation in the background could be stimulated anywhere from 10-fold to more genotype of the strains used. than 100-fold depending on the configuration of the Together radlA and rad52-8 interactsynergistically double-strand break site and the recombining alleles. to decrease recombination in the presence or absence THOMASand ROTHSTEIN(1989a) found that only of HOTl. Thisindicates that these two genes function plasmid excision was stimulated when high levels of in different recombination pathways that can compete RNA polymerase I1 transcription occurred. The dif- for at least some of the same intermediates. A similar ference between the results reported here and those observation has been made by SCHIESTLand PRAKASH of THOMASand ROTHSTEIN(1 989a) may be that RNA (1988), KLEIN (1988) and THOMASand ROTHSTEIN polymerase I transcription (from HOTl ) stimulates (198913). HOTI stimulates the rate of gene replace- recombination by a different mechanism than RNA mentbut not plasmid excision in radlA rad52-8 polymerase I1 transcription. Alternatively, this differ- strains. Thus a portion of the activity of HOTl occurs ence may result from the use of different recombina- by a RADl- and RAD52-independent pathway. tion substrates. Mutations in RAD4 and RAD50 affect interchro- The other forms of HOTl, BglII-B HOTl andIE t mosomal mitotic recombination (MONTELONE, HOEK- HOTI, also affect recombination in this assay system STRA and MALONE1988; MALONEand ESPOSITO198 1 ; in wild-type strains. As found previously BglII-B HOTl MALONE 1983). In our intrachromosomal assay, re- produces a much smaller increase in recombination gardless of the presence or absence of HOTl, these than E1 HOTl (VOELKEL-MEIMAN,KEIL and ROE- mutations did not alter the rateof recombination. In DER 1987). The IE c HOTl construct that is inactive light of the synergistic interaction of radl andrad52, in stimulating the rateof recombination (Table 5 and studies examining the interaction of these rad muta- VOELKEL-MEIMAN,KEIL and ROEDER1987) doeshave tions with radl and rad52 are warranted. However, a subtle effect on recombination in that a significantly it is clear that interchromosomal and intrachromoso- smaller percentage of His+ recombinants are gene mal exchange are affected differently by some muta- replacements (Table 4). Thus the presence of the IE tions involved in therepair and recombination of -+HOTl insertion may affect the resolution of recom- DNA. For example, rad50 increases spontaneous mi- bination intermediates or the types of recombination totic interchromosomal recombination three- to ten- events that occur although it does not alter the rate fold (MALONEand ESPOSITO1981 ; MALONE 1983) of recombination. but, as shown in Table 1, does not affect intrachro- Effects of rad mutations on the recombination- mosomal exchange. stimulatory activity of HOTl: Rad1A significantly His+ Ura+ diploids in rud52-8 strains: The most 50 B. R. Zehfus et al. striking observation in rad52-8 strainsis that His+Ura+ OLOFF et al. 1989)during mitosis. It has notbeen recombinants are not products of gene replacement shown whether RAD52 is also required for the repair but rather arediploids apparently derived by endom- of other lesionsin DNA but RESNICKet al. (1984) itosis and excisive recombination. This occurs in the found that single-strand nicks accumulate in rad52 presence or absence of HOTl. There areseveral pos- cells during meiosis. Our findingthat part of the sible explanations for this observation. One is that HOTl activity is RAD52-dependent suggests that some recombination and endomitosis are associated events of these lesions may be double-strand breaks, may be in these cells. Possibly recombination is initiated but processed into double-strand breaks or may be other cannot be properly resolved. The cells may undergo lesions whose repair requires RAD52. Using an inter- an abortive mitosis without nuclear division, followed chromosomal recombination assay for HOT1 activity, by resolution of the recombination intermediate. An VOELKEL-MEIMANand ROEDER(1 990) found that the alternative explanation is that in rad52-8 strains en- chromosomecontaining HOTl is preferentially the domitosis results in the production of diploid cells that recipient of information during geneconversion. This then can undergo intrachromosomal recombination is predicted for recombinationinitiated by double- on one chromosome. This results in the production strandbreaks (SZOSTAKet al. 1983).Studying the of a diploid cell with one His+Ura- chromosome and effects of double-strand breaks on mitotic exchange one His-Ura+ chromosome. The rate of recombina- of direct repeats, J. A. NICKOLOFFand F. HEFFRON tion to produce a His+Ura- chromosome in a rad52- (personal communication) found thatthe HO cleavage 8 strain in the absence of HOTl is 0.23 X (Table site is absent in recombinantsrecovered following 5). If preexisting diploids in the rad52-8 culture pro- HO-nuclease induced recombination. Thus the HO- duce the His+Ura+ diploids, one-seventh of the cells double-strand break site is always lost during recom- in the culture would have to be diploid to produce bination. Since we find that HOTl is frequently re- the observed His+Ura+ diploid rate of 0.07 X covered in recombinants(Table 6), it appearsthat [that is (0.07 X 10-'))/(2 X (0.23 X lo-'))]. Experi- HOTl need not be the site of a double-strand break. ments arecurrently in progressto determine the Rather, HOTl may increase the frequency of double- frequency of diploid cells present in cultures of rad52- strandbreaks in adjacentDNA, such as the direct 8 strains. repeats or the intervening plasmid sequences. NICK- Other examples of genomic instability have been OLOFF et al. (1989) found that double-strand breaks reported in rad52 cells. MORTIMER, CONTOPOULOU in either of these regions stimulated both gene con- and SCHILD (1981)showed that there is a high fre- version and excisive exchange. THOMASand ROTH- quency of chromosome loss during mitosis in diploids STEIN (1989a) proposed that RNA polymerase I1 tran- homozygous forrad52. HABER andHEARN (1985) scription stimulates recombination by increasing the found thatmitotic interchromosomal geneconversion occurrence of double-strandbreaks in the plasmid in rad52 strains oftenis accompanied by chromosome sequences separating the direct repeats. loss. THOMASand ROTHSTEIN(1 98913) recovered cells RADl is required for the incision of UV-damaged that were disomic forthe chromosome containing DNA (REYNOLDSand FRIEDBURC 1981;WILCOX and theirrecombination substrate among recombinants PRAKASH198 1). Since UV-excision repair involves the fromrad52-8 strains. L.-W. YUAN and R. L. KEIL formation andthen repair of a single-strand gap, (unpublished results) have observed both diploid and RADl may be involved in the formation or repair of disomic recombinants in another rad52-8 strain. similar lesions during recombination. SCHIESTLand Mechanism of recombination stimulation by PRAKASH (1988)previously proposed such a role for HOTZ: The recombination-stimulatory activity of RADl in recombination. Rad1A has a significant effect HOTlappears to depend on the ability of this se- on the rate of excisive recombination when HOTl is quence to initiate high levelsof transcription present but not when it is absent for this direct-repeat (VOELKEL-MEIMAN,KEIL and ROEDER1987; STEW- recombination assay. Therefore, instead of uniformly ART and ROEDER1989). This transcription may per- increasing the occurrenceof all lesions, HOTl appears mit the introduction of lesions in the DNA that stim- to preferentially increase theproportion of lesions ulate recombination. Possible lesions include single- that are processed by the RADI-dependent pathway. stranded regions of DNA, nicks, breaks or RNA:DNA The synergistic interaction of radlA and rad52-8 in- hybrids as suggested by ROSENBERG(1988). RAD52 dicates that some lesions can be processed by either has been shown to be required for the recombina- recombination pathway. tional repair of double-strand breaks in mating-type Instead of allowing more frequent introduction of switching (MALONE and ESPOSITO1980; WEIFFEN- lesions in DNA, HOTl could stimulate recombination BACH and HABER198 1; STRATHERNet al. 1982), yeast by affecting another stepin exchange such as pairing, transformation(ORR-WEAVER, SZOSTAK and ROTH- conversion-tract length, or resolution. AHN and LIV- STEIN 198l), and homologous recombination (NICK- INGSTON (1986)found that the average length of H OT I-Stimulated Recombination HOTI-Stimulated 51 conversion for mitotic recombination is 0.5 kb. If by HOTl athis4 (Y.-H. LINand R. L. KEIL, manuscript HOTl increased the conversion-tractlength, more in preparation). recombination events that initiated at a distance from We thank ANITAK. HOPPER, PAUL SZAUTERand two anonymous the his4-260 mutation would include this site and reviewers for their thoughtful comments regarding the manuscript. increase the frequency of His+ recombinants. How- We thank R.E. MALONEfor providing space for M.F.H. during the ever, our data indicatethat HOTl doesnot affect early stages of this work. M.F.H. is a Lucille P. Markey scholar in conversion-tract length. In wild-type strains, the pro- Biomedical Sciences and is supported by a grant from the Markey portion of His+ Ura+ gene convertantsin whichHOTl CharitableTrust. This work was funded by U.S. PublicHealth Service grant GM36422 from the National Institutes of Health to is coconverted with the wild-type his4-260 sequence R.L.K. is not significantly different from the proportion of coconversion for the BamHI linker and the wild-type LITERATURE CITED his4-260 sequence (Table 6). HOTl also doesnot AHN,B.-Y., and D. M. LIVINGSTON,1986 Mitotic gene conversion appear to introduce a preferential resolution site for lengths, coconversion patterns, and the incidence of reciprocal exchange. If, for example, HOTl introduced a reso- recombination in Saccharomycesa cereuisiae plasmid system.Mol. lution site that usually produced reciprocal exchange, Cell. Biol. 6 3685-3693. it could increase the rate of plasmid excision. Such a ALAM,S. N., T. K. SHIRESand H. Y. ABOUL-ENEIN,1975 An site would alter the position of exchanges. We find improved synthesis andmass fragmentometry of 5-fluoroorotic acid. Acta Pharm. Suec. 12: 375-378. that in wild-type strains E1 + HOTl only slightly BAYEV,A. A,, 0. I. GEORGIEV,A. A. HADJIOLOV,M. B. KERMEK- affects the position of exchange in excisive events as CHIEV, N. NIKOLAEV,K. G. SKRYABIN andV. M. ZAKHARYEV, judged by the proportion of Ura- recombinants that 1980 The structure ofthe yeastribosomal RNA genes. 2. are His- (Table 4). More precise localization of the The nucleotide sequence of the initiation site for ribosomal RNA transcription. Nucleic Acids Res. 8: 4919-4926. position of the exchange by examining independent BLACKWELL,T. K., M. W. MOORE,G. D. YANCOPOULOS, H. SUH, Ura- recombinants for the presence or absence of S. LUTZKER,E. SELSINGand F. W. ALT, 1986 Recombination HOTl or theBamHI linker (in strains lacking HOTl) betweenimmunoglobulin variable region gene segments is also shows nodramatic difference between strains enhanced by transcription. Nature 324 585-589. containing and lacking HOTl (Table 6). DRAKE, J.W., 1970 The Molecular Basis of Mutation. Holden-Day, San Francisco. Mutations in RAD52 do not affect rDNA recombi- ELION, E. A., andJ. R. WARNER,1984 The majorpromoter nation (ZAMB and PETES 1981; PRAKASHand TAIL- element of rRNA transcription in yeastlies 2 kb upstream.Cell LON-MILLER198 1; R. L. KEIL, Y.-H. LIN and A. D. 39663-673. MCWILLIAMS,unpublished results). Recombination in ELION, E. A., and J. R.WARNER, 1986 An RNApolymerase I the rDNA is not decreased in radlA or radlArad52- enhancer in Saccharomyces cerevisiae. Mol. Cell. Biol. 6: 2089- 2097. 8 strains(R. L. KEIL, Y.-H. LIN and A. D. Mc- GAME,J. C., T. J. ZAMB, R. J. BRAUN,M. RESNICK and R. M. ROTH, WILLIAMS,unpublished results). Why then is HOTl 1980 The role of radiation (rad) genes in meiotic recombi- activity decreased by these mutations? It is possible nation in yeast. Genetics 94 51-68. that HOTl andRADlIRAD52 are involved in different HABER,J. E., and M. HEARN,1985 rad524ndependentmitotic gene conversion inSaccharomyces cerevisiae frequently resultsin parts of the recombination pathway. Forexample, chromosomal loss. Genetics 111: 7-22. HOTl may affect initiation of exchange while RADl HAYNES, R. H., andB. A. KUNZ,1981 DNA repair and mutagen- and RAD52 are involved in a later step such as reso- esis in yeast, pp. 371-414 in The Molecular Bioloy of the Yeast lution. When HOTl is inserted at HZS4, an exchange Saccharomyces, edited by J. N. STRATHERN,E. W. JONES and J. event may beinitiated in HOTland then branch R. BROACH.Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. migrate into adjacent non-rDNAsequences. Once the JACKSON, J. A., and G. R. FINK, 1981 Gene conversion between recombinationintermediate leaves HOTl,the ex- duplicated elements in yeast. Nature 292: 306-3 1 1. change may be dependent on all of the machinery KEIL, R. L., andG. S. ROEDER,1984 Cis-actingrecombination- normally required for non-rDNA recombination in- stimulating activity in a fragment of the ribosomal DNA of S. cluding RADl and RAD52. Thus the recombination- cereuisiae. Cell 39: 377-386. KEMPERS-VEENSTRA,A. E., J. OLIEMANS,H. OFFENBERG,A. F. stimulatory activity of HOTl would be reduced by DEKKER,R. J. PLANTAand J. KLOOTWIJK,1986 3’-end forma- mutations in RADl and/or RAD52. Alternatively, it is tion of transcripts from the yeast rRNA operon. EMBO J. 5: possible that therDNA, which is contained in the 2703-27 10. nucleolus, is inaccessible tothe gene products re- KLAR, A. J. S., J. N.STRATHERN and J. A.ABRAHAM, 1984 Involvement of double-strand chromosomal breaks for quired for the RADl- and RAD52-dependent recom- mating-type switching in Saccharomyces cerevisiae. Cold Spring bination pathways. These findings do not mean that Harbor Symp. Quant. Biol. 49: 77-88. HOTl doesnot function in rDNA recombination. KLEIN, H. L., 1988 Differenttypes of recombination events are There may beother factors thatare required for controlled by the RADl and RAD52 genes of Saccharomyces HOTl activity both at his4 and in the rDNA. We have cerevisiae. Genetics 120 367-377. KLEMENZ,R., and E.P. GEIDUSCHEK,1980 The 5’ terminusof recently isolated mutationsthat affect bothrDNA the precursor ribosomal RNA of Saccharomyces cereuisiae. Nu- recombination and the stimulation of recombination cleic Acids Res. 8: 2679-2689. 52 B. R. Zehfus et al.

LEA,D. E., and C. A. COULSON,1949 The distribution of the ROTHSTEIN,R. J., 1983 One-step gene disruption in yeast. Meth- numbers of mutants in bacterialpopulations. J. Genet. 49 ods Enzymol. 101: 202-21 1. “4-285. SCHIESTL,R. H., and S. PRAKASH,1988 RADI, an excision repair MALONE,K. E., 1Wl Multiple mutant analysis of recombination gene of Saccharomyces cereuisiae is also involved in recombina- in yeast. Mol. Gen. Genet. 189: 405-412. tion. Mol. Cell. Biol. 8: 3619-3626. MAI.ONE, R. E., and R. E. ESPOSITO,1980 The RAD52 gene is SCHLISSEL,M. S., and D. BALTIMORE,1989 Activationof inmu- required for homothallic interconversion of mating types and noglobulin kappa gene rearrangement correlates with induc- spontaneous mitotic recombination in yeast. Proc. Natl. Acad. tion of germline kappa gene transcription.Cell 58: 100 1- 1007. Sci. USA 77: 503-507. SHERMAN,F., G. R. FINKand J. B. HICKS, 1986 Methodsin Yeast MALONE,K. E., and K. E. ESPOSITO,1981 Recombinationless Genetics. Cold Spring Harbor Laboratory,Cold Spring Harbor, meiosis in Saccharomyces cereuisiae. Mol. Cell. Biol. 1: 891-901. N.Y. MALONE,K. E., J. E. GOLINand M. S. ESPOSITO,1980 Mitotic SIEGEL,S., 1956 Nonparametric Statisticsfor the Behavioral Sciences. versus meiotic recombination in Saccharomyces cereuisiae. Curr. McGraw-Hill, New York. Genet. 1: 241-248. SNEDECOR,G. W., and W. G. COCHRAN,1967 StatisticalMethods, MANIATIS,T., E. E. FRITSCHand J. SAMBROOK,1982 Molecular Ed. 6. Iowa State University Press, Ames. Cloning: A LaboratoryManual. Cold Spring Harbor Laboratory, STEWART,S. E., and G. S. ROEDER,1989 Transcription by RNA Cold Spring Harbor, N.Y. polymerase 1 stimulates mitotic recombination in Saccharomyces MONTELONE, B. A,, M. F. HOEKSTRAand R. E. MALONE, cereuisiae. Mol. Cell. Biol. 9 3464-3472. 1988 Spontaneous mitotic recombination in yeast: the hyper- STRATHERN,J. N., A. J. S. KLAR,J. B. HICKS,J. A. ABRAHAM,J. M. recombinational rem1 mutations are alleles of the RAD3 gene. IVY, K. A. NASMYTHand C. MCGILL, 1982Homothallic Genetics 119: 289-301. switching of yeast mating typecassettes is initiated by a double- MORTIMER,R. K., R. CONTOPOULOUand D. SCHILD,1982 Mitotic stranded cut in the MAT locus. Cell 31: 183-192. chromosome loss in aradiation-sensitive strain of the yeast SUN, H., D. TRECO, N.P. SCHULTESand J. W. SZOSTAK, Saccharomyces cereuisiae. Proc. Natl. Acad. Sci. USA 78: 5778- 1989 Double-strand breaksat an initiationsite for meiotic 5782. gene conversion. Nature338: 87-90. NICOLAS, A,, D. TRECO,N. P. SCHULTESand J. W. SZOSTAK, SZOSTAK,J. W., T. L. ORR-WEAVER,R. J. ROTHSTEINand F. W. 1989 An initiationsite for meiotic gene conversion in the STAHL,1983 The double-strand-breakrepair model for re- yeast Saccharomyces cereuisiae. Nature 338: 35-39. combination. Cell 33: 25-35. NICKOLOFF,J. A., J. D. SINGER,M. F. HOEKSTRAand F. HEFFRON, THOMAS,B. J., and R. ROTHSTEIN,1989a Elevated recombination 1989 Double-strand breaks stimulate alternative mechanisms rates in transcriptionally active DNA. Cell 56: 619-630. of recombination repair. J. Mol. Biol. 207: 527-541. THOMAS,B. J., and R. ROTHSTEIN,1989b The genetic control of ORR-WEAVER,T. L., J. W. SZOSTAKand R. J. ROTHSTEIN, direct-repeatrecombination in Saccharomyces: the effectof 1981 Yeast transformation: a model system for the study of rad52 and rad1 on mitotic recombination at GALIO, a tran- recombination. Proc. Natl. Acad. Sci. USA 78: 6354-6358. scriptionally regulated gene. Genetics 123: 725-738. PRAKASH,I.., and P. TAILLON-MILLER, 1981Effects of the rad52 VOELKEL-MEIMAN,K., R. L. KEIL and G. S. ROEDER, gene on sister chromatid recombination in Saccharomyces cere- 1987 Recombination-stimulatingsequences in yeast ribo- uisiae. Curr. Genet. 3: 247-250. somal DNA correspond to sequences regulating transcription PRAKASH,S., L. PRAKASH,W. BURKEand B. A. MONTELONE, by RNA polymerase I. Cell 48 1071-1079. 1980 Effects of the RAD52 gene on recombination in Saccha- VOELKEL-MEIMAN,K., andG. S. ROEDER,1990 A chromosome romyces cereuisiae. Genetics 94: 3 1-50. containing HOT1 preferentially receives informationduring RESNICK,M. A,,and P. MARTIN, 1976 The repairof double- mitotic interchromosomal gene conversion. Genetics124: 56 1 - strand breaks in the nuclear DNA of Saccharomyces cereuisiae 572. and its genetic control. Mol. Gen. Genet. 143: 119-129. WEIFFENBACH,B., and J. HABER, 1981 Homothallic mating type RESNICK, A,, T. CHOW,J. NITISSand J. GAME,1984 Changes M. switching generates lethal chromosome breaks in rad52 strains in the chromosomal DNA of yeast during meiosis in repair of Saccharomyces cereuisiae. Mol. Cell Biol. 1: 522-534. mutants and the possible role of a deoxyribonuclease. Cold WILCOX,D. R., and L. PRAKASH,1981 lncision and postincision Spring Harbor Synlp. Quant. Biol. 49 639-649. steps of pyrimidine dimer removal in excision-defective mu- REYNOLDS,R. J., and E. C. FRIEDBERG,1981 Molecularmecha- tants of Saccharomyces cereuisiae. J. Bacteriol. 143: 61 8-623. nisnls of pyrimidine dimer excision in Saccharomyces cerevisiae: YANCOPOULOS,G. D., T. K. BLACKWELL, H. SUH,L. HOOD and F. incision of ultraviolet-irradiated deoxyribonucleic acid in vivo. W. ALT, 1986Introduced T cell receptor variable region J. Bacteriol. 146: 692-704. gene segments recombine in pre-B cells: evidence that B and RONNE,H., and R. ROTHSTEIN,1988 Mitotic sectored colonies: T cells use a common recombinase. Cell 44 251-259. evidence of heteroduplexDNA formation during direct repeat YUAN,L.-W., andR. L. KEIL, 1990Distance-independence of recombination. Proc. Natl. Acad. Sci. USA 85: 2696-2700. mitotic intrachromosomal recombination in Saccharomyces cere- ROSE, M., and F. WINSTON,1984 Identification of a Ty insertion uisiae. Genetics 124: 263-273. within the coding sequenceof the S. cereuisiae URA3 gene. Mol. ZAMB,T. J., and T. D. PETES,1981 Unequal sister-strand recon]- Gen. Genet. 193: 557-560. bination within yeast ribosomal DNAdoes not require the ROSENBERG,S. M., 1988 Chain-bias of Escherichia coli rec-me- RAD52 gene product. Curr. Genet. 3: 125-132. diated X patch recombinants is independent of the orientation of X cos. Genetics 119: 7-2 1. Communicating editor: D. BOTSTEIN