Genetic Analysis of Yeast RPA1 Reveals Its Multiple Functions in DNA Metabolism

Corresponding author: Richard D. Kolodner, Ludwig Institute for Cancer Research, UC San Diego School of Medicine, CMME-3080, 9500 Gilman Drive, La Jolla, CA 92093-0660, rkolodner{at}ucsd.edu (E-mail).

Communicating editor: L. S. SYMINGTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Replication protein A (RPA) is a single-stranded DNA-binding protein identified as an essential factor for SV40 DNA replication in vitro. To understand the in vivo functions of RPA, we mutagenized the Saccharomyces cerevisiae RFA1 gene and identified 19 ultraviolet light (UV) irradiation- and methyl methane sulfonate (MMS)-sensitive mutants and 5 temperature-sensitive mutants. The UV- and MMS-sensitive mutants showed up to 10⁴ to 10⁵ times increased sensitivity to these agents. Some of the UV- and MMS-sensitive mutants were killed by an HO-induced double-strand break at MAT. Physical analysis of recombination in one UV- and MMS-sensitive rfa1 mutant demonstrated that it was defective for mating type switching and single-strand annealing recombination. Two temperature-sensitive mutants were characterized in detail, and at the restrictive temperature were found to have an arrest phenotype and DNA content indicative of incomplete DNA replication. DNA sequence analysis indicated that most of the mutations altered amino acids that were conserved between yeast, human, and Xenopus RPA1. Taken together, we conclude that RPA1 has multiple roles in vivo and functions in DNA replication, repair, and recombination, like the single-stranded DNA-binding proteins of bacteria and phages.

REPLICATION Protein A [RPA, also known as Replication Factor A (RFA) or human SSB],³ is a single-stranded DNA- (ssDNA-) binding protein that was initially identified as an essential factor for replication of SV40-ori containing DNA in vitro and is likely to be the eukaryotic equivalent of the Escherichia coli SSB protein required for most aspects of bacterial DNA metabolism (STILLMAN 1989 ; TSURIMOTO et al. 1989; KENNY et al. 1990 ; COLLINS and KELLY 1991 ; LEE et al. 1991 ; DORNREITER et al. 1992 ; BOCHKAREV et al. 1997 ). Numerous biochemical studies have supported this point of view. Analysis of RPA in the SV40 replication system revealed that it is involved in both the initiation and elongation phases of DNA replication (STILLMAN 1989 ; TSURIMOTO and STILLMAN 1989 ; KENNY et al. 1990 ; COLLINS and KELLY 1991 ; LEE et al. 1991 ; DORNREITER et al. 1992 ). RPA is also involved in other aspects of DNA metabolism. RPA is required for in vitro reconstitution of nucleotide excision repair (COVERLEY et al. 1991 ; ABOUSSEKHRA et al. 1995 ; KAZANTSEV et al. 1996 ) and its physical association with and functional stimulation of XPA, XPG, and XPF, along with its ability to bind to UV-damaged DNA, further confirm the importance of RPA in nucleotide excision repair (HE et al. 1995 ; LEE et al. 1995 ; LI et al. 1995 ; MATSUDA et al. 1995 ; BURNS et al. 1996 ; MATSUNAGA et al. 1996 ). Antibody depletion and in vitro reconstitution studies have demonstrated that RPA is required for mismatch repair (LIN et al. 1997). Finally, RPA stimulates the in vitro strand exchange activities of Saccharomyces cerevisiae SEP1 (Strand Exchange Protein), human HPP1 (Homologous Pairing Protein), and RAD51 (a eukaryotic homologue of the E. coli RecA) (HEYER et al. 1990 ; MOORE et al. 1991 ; ALANI et al. 1992 ; SHINOHARA et al. 1992 ; OGAWA et al. 1993 ; SUNG 1994 , SUNG 1997 ; SUGIYAMA et al. 1997 ), and human RPA physically interacts with the human RAD52 gene product via RPA2 (PARK et al. 1996 ) consistent with a role in genetic recombination. In sum, good biochemical evidence exists that RPA plays a crucial role in DNA replication, repair, and recombination.

RPA is an evolutionarily conserved protein consisting of three subunits: RPA1 (Mr 70,000), RPA2 (Mr 36,000), and RPA3 (Mr 14,000). Of these three subunits, RPA1 is the best characterized biochemically. While most biochemical characterization of RPA1 has been done using human RPA1, extrapolation of these results to other eukaryotic RPAs seems justified given the high degree of conservation among the various RPA homologues (HEYER et al. 1990 ; BRILL and STILLMAN 1991 ; ERDILE et al. 1991 ; ADACHI and LAEMMLI 1992 ; BROWN et al. 1994 ; ISHIAI et al. 1996 ). Functional dissection of human RPA1 has identified three distinct domains. The N-terminal 100 amino acids of human RPA1 are required for stimulation of DNA polymerase alpha (KIM et al. 1996 ). The ssDNA-binding activity of human RPA1 resides between amino acid residues 175–420 (GOMES and WOLD 1995 ; KIM et al. 1996 ; LIN et al. 1996 ). X-ray crystallographic analysis of this region suggested that it is comprised of two structurally homologous subdomains oriented in tandem (BOCHKAREV et al. 1997 ). The overall structure of the RPA1 subdomains appears similar to those of previously solved ssDNA-binding proteins, T4 gp32 and f1 gene V protein (BOCHKAREV et al. 1997 ). The C-terminal 100 amino acids of human RPA1 are required for binding RPA2 and RPA3 to form the RPA heterotrimer (GOMES and WOLD 1996 ; LIN et al. 1996 ). Located within the C-terminal third of RPA1 is a putative C4 zinc finger motif conserved among all eukaryotic homologues (ERDILE et al. 1991 ; ADACHI and LAEMMLI 1992 ; BROWN et al. 1994 ; ISHIAI et al. 1996 ), and this region of RPA1 appears to be required for DNA replication and mismatch repair but not nucleotide excision repair (LIN et al. 1996 ; LIN et al. 1997).

While RPA1 alone binds to ssDNA with high affinity (BRILL and STILLMAN 1989 ; WOLD et al. 1989 ; KENNY et al. 1990 ), the complete holocomplex is required for DNA replication in vitro (ERDILE et al. 1991 ). Consistent with this observation, all three subunits of RPA are required for viability in S. cerevisiae (HEYER et al. 1990 ; BRILL and STILLMAN 1991 ). Although the functions of RPA2 and RPA3 remain to be elucidated, recently it was shown that the minimal regions of RPA2 and RPA3 necessary to maintain viability in S. cerevisiae bear weak sequence similarity to the E. coli SSB and to the two ssDNA-binding subdomains of RPA1. Based on this observation, it was proposed that the four ssDNA-binding subdomains of RPA function like the homotetrameric E. coli SSB (MANIAR et al. 1997 ). Consistent with this hypothesis is the observation that S. cerevisiae RPA2 has contacts with ssDNA (PHILIPOVA et al. 1996 ). Also consistent is the isolation of S. cerevisiae RFA2 and RFA3 alleles that are defective in DNA replication (SANTOCANALE et al. 1995 ; MANIAR et al. 1997 ). In addition to a role in ssDNA binding, RPA2 has been implicated as the regulatory subunit of the RPA heterotrimer (DIN et al. 1990 ; DUTTA and STILLMAN 1992 ; FOTEDAR and ROBERTS 1992 ; CARDOSO et al. 1993 ; LIU and WEAVER 1993 ; CARTY et al. 1994 ).

While there is considerable biochemical data suggesting that RPA functions in a variety of DNA metabolic processes, only limited genetic analysis has been performed to elucidate its roles in vivo. Initial studies using rfa1 null alleles reported that RFA1 is an essential gene and demonstrated that rfa1 mutations caused a terminal arrest phenotype consistent with a defect in DNA replication (HEYER et al. 1990 ; BRILL and STILLMAN 1991 ). One subsequent study reported the construction of two-amino-acid insertion mutations in RFA1, some of which caused weak temperature sensitivity, weak UV-sensitivity, and a modest defect in mitotic intrachromosomal recombination (LONGHESE et al. 1994 ). Another study reported an rfa1 mutation (rfa1-44) that caused a defect in HO-endonuclease-induced plasmid-to-chromosome gene conversion and sensitivity to UV and X rays (FIRMENICH et al. 1995 ). Overexpression of RAD52 suppressed the rfa1-44 mutation, providing evidence that RPA and RAD52 interact (FIRMENICH et al. 1995 ). Subsequent studies of mating type switching using physical assays of recombination detected only an apparant 50% failure in recombination caused by the rfa1-44 allele (N. SUGAWARA and J. E. HABER, unpublished data). A third rfa1 mutation was isolated in a screen for suppressors of the defect in direct repeat recombination in rad1 rad52 double mutants, and this rfa1 allele by itself was found to cause some defects in genetic recombination (SMITH and ROTHSTEIN 1995 ). Other studies of RFA1 have suggested a role for RPA in checkpoint control (BRUSH et al. 1996 ; LONGHESE et al. 1996 ; PARKER et al. 1997 ) and have suggested that RPA could serve as a DNA damage sensor (BRUSH et al. 1996 ).

While previous studies described the first rfa1 mutants, they either screened for a single phenotype and identified a single mutant (FIRMENICH et al. 1995 ; SMITH and ROTHSTEIN 1995 ), or they isolated a small number of mutants and then tested them for significant phenotypes (HEYER et al. 1990 ; BRILL and STILLMAN 1991 ; LONGHESE et al. 1994 ). Consequently, they may not have been capable of revealing the wide variety of metabolic defects that could be caused by mutations in the RFA1 gene. In the present study, we focused on randomly mutagenizing the RFA1 gene and systematically screened for mutations that confer a temperature-sensitive (ts) and UV- or MMS-sensitive phenotype in order to isolate a variety of alleles. We isolated 24 rfa1 alleles and characterized 19 DNA-damage-sensitive mutants and two ts mutants in greater detail.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media:
E. coli strains were grown in luria bertani (LB) medium (MILLER 1972 ), which was supplemented with 100 µg/ml ampicillin when required. SOC medium (SAMBROOK et al. 1989 ) was used for incubation of the cells after electro/transformation. Complex glucose [yeast extract reptone adenine dextrose (YPD)], sporulation, synthetic complete (SC), and various drop-out media for growth of yeast strains were made as previously described (SHERMAN et al. 1983 ). Adenine sulfate was added to YPD at 0.004% (YPAD) when indicated. 5-Fluoro-orotic acid (5-FOA) was used as described (ROSE et al. 1990 ). Plates containing Methyl Methane Sulfonate (MMS, Kodak, Rochester, NY) were used within two days after preparing them. For HO-induction experiments designed to test for sensitivity to HO expression, cells were grown in SC medium lacking glucose, uracil, and leucine and containing 3% glycerol and 2% sodium lactate (pH 5.5). To induce the GAL-HO gene, galactose (2%) was added to liquid cell cultures grown in this medium, or the culture was spotted onto solid medium lacking glucose, uracil, and leucine and containing 2% galactose. For analysis of HO-induced mating type switching, cells were propagated in yeast extract peptone-lactate medium and induced with galactose as previously described (CONNOLLY et al. 1988 ).

S. cerevisiae and E. coli strains:
Strain YSB89 (MATa/MAT ade2-1/ade2-1 ura3-1/ura3-1 his3-11,15/his3-11,15 trp1-1/trp1-1 leu2-3,112/leu2-3,112 can1-100/can1-100 rfa1::TRP1/RFA1) derived from the parental strain W303 was kindly provided by Dr. B. STILLMAN (Cold Spring Harbor Laboratory) (BRILL and STILLMAN 1991 ). Strains RKY 2102 [MATa ade2-1 ura3-1 his3-11 trp1-1 leu2-3,112 can1-100 rfa1::TRP1 (pRPA1 URA3 RFA1)] and RKY 1900 [MATa ade2-1 ura3-1 his3-11 trp1-1 leu2-3,112 can1-100 rfa1::TRP1 (pKU1 LEU2 RFA1)] were constructed by transforming YSB89 with plasmids pRPA1 and pKU1 (see below), respectively, sporulating the strain, and then identifying an appropriate spore clone after tetrad dissection. The isogenic wild-type strain W303-1A was obtained as a spore clone from YSB89. To obtain the strain used for analysis of single-strand annealing, the W303 strain containing the rfa1-t11 allele was crossed to tNS1373M [ho HML leu2 MAT-inc hmr-3 mal2 trp1 thr4 GAL+ ura3-Nco-pUC-HOcs-H3(2.3 kb)-URA3 GAL::HO at THR4] (SUGAWARA and HABER 1992 ) to obtain rfa1-t11 segregants that were ura3 leu2 mat -inc and possessed GAL-HO integrated at the THR4 locus. The resulting strain was then transformed with pNSU208 (see below). The E. coli strains DH5 and DH5 were used to amplify and manipulate all plasmids described in this study.

Plasmids:
The plasmid pRPA1 (HEYER et al. 1990 ) contains a 3.4-kb Sau3A insert carrying the RFA1 gene in the BamHI site of the ARS1 CEN4 URA3 vector YCp50. Plasmid pKU1 was constructed by subcloning the 3.1-kb Sal I-HindIII fragment containing the RFA1 gene from the plasmid pRPA1 into the corresponding sites of the ARSH4 CEN6 LEU2 vector pRS415. A series of plasmids for rfa1 allele replacement pKU2 was constructed by inserting the 3.1-kb Sal I-BamHI fragment of pKU1 derivatives carrying the rfa1 mutant alleles into the corresponding sites of the URA3 integration plasmid pRS406. Plasmids pKU1 and pKU2 carrying the rfa1 mutants are indicated by hyphenating allele names. The GAL-HO plasmids pJH132 (GAL-HO URA3 ARS1 CEN4) and pJH727 (GAL1-HO LEU2 ARS1 CEN4) have been described ( JENSEN and HERSKOWITZ 1984 ; WHITE and HABER 1990 ). The plasmid pNSU208 was constructed by inserting the EcoRV-BssHII fragment from the lacZ sequence of pJF3 into the HindIII site of pNR16 (FISHMAN-LOBELL et al. 1992 ).

Genetic techniques:
E. coli strains were transformed with DNA by electroporation using Gene Pulser (Bio-Rad, Richmond, CA). Standard methods for yeast genetics were essentially according to ROSE et al. 1990 . Transformation of yeast cells was performed using the lithium acetate procedure as described (GIETZ et al. 1992 ). Plasmid DNA was isolated as described (ROSE et al. 1990 ). Allele replacement was performed by standard methods involving transformation with pKU2 series plasmids that had been linearized at a site in the RFA1 gene by digestion with NheI.

RFA1 mutagenesis:
Mutagenesis of the RFA1 gene was performed by PCR based on the technique of MUHLRAD et al. 1992 , except that standard PCR conditions were used. The error rate of Taq polymerase is in the range of 0.1 x 10^-4 to 2 x 10^-4 per nucleotide under standard PCR conditions (CADWELL and JOYCE 1994 ), and at this error rate we estimated that 12 cycles of standard PCR over a region of about 2 kb of the RFA1 open reading frame (ORF) was sufficient to generate a diverse mutation library. PCR primers were designed to amplify the 2.8-kb region of pKU1 containing the RFA1 open reading frame and the flanking sequences on both sides of the open reading frame. The primers 5'-ATGGTGCATGCAAGGAGATGGC and 5'-CACTTTATGCTTCCGGCTCCTA, which were synthesized by Molecular Biology Core Facility, Dana-Farber Cancer Institute, Boston, MA, were complementary to vector sequences located 180 nucleotides upstream and downstream from the unique MluI and HindIII sites in the vector, respectively. The MluI and HindIII sites are located 165 bp upstream from the initiation codon and 252 bp downstream from the termination codon of the RFA1 open reading frame, respectively. PCR mixtures (25 µl) contained 1.25 units of Taq DNA polymerase (Perkin Elmer, Norwalk, CT), 10 ng of template DNA (pKU1 linearized at the unique ClaI site in the LEU2 gene), 1 µM of each primer, 200 µM each dNTP, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl₂, and 0.001% (w/v) gelatin. PCR was performed in a Perkin Elmer 9600 utilizing an initial incubation at 95° for 5 min, followed by 12 cycles of 30 sec at 95°, 30 sec at 45°, and 90 sec at 72°. Twenty independent reaction mixtures were combined (total 500 µl) and purified using a QIAquick Spin Column (QIAGEN, Santa Clarita, CA). The PCR product and the 6.8-kb MluI-HindIII fragment of pKU1 (the vector fragment of pKU1 obtained by digestion with MluI-HindIII, followed by purification by agarose gel electrophoresis to remove the 2.4-kb RFA1-containing fragment) were mixed in a 1:1 molar ratio and cotransformed into RKY 2102. Transformants were selected on SC plates lacking Leu and containing Ura and then replica-plated onto 5-FOA plates to eliminate pRPA1 by plasmid shuffling (BOEKE et al. 1987 ). These 5-FOA plates were used as master plates for screening of rfa1 mutants. Selection and growth of transformants during this mutagenesis procedure were at 26°.

Screening of rfa1 mutants:
To screen candidate rfa1 mutants for those having temperature-sensitive (ts) growth, UV-sensitive (UV^S), and MMS-sensitive (MMS^S) phenotypes, each 5-FOA master plate was replica-plated onto a set of four YPD plates; 7.5 µM of erythrosine B (Sigma, St. Louis, MO) was included in the plates to stain nonviable cells (BONNEU et al. 1991 ). In addition, MMS (0.01%) was included in one of these four plates. In some cases, adenine sulfate (YPAD) was added to suppress coloring due to the ade2 mutation and to make the red staining by erythrosine B more visible (BONNEU et al. 1991 ). One of the YPD (or YPAD) plates was irradiated with 50 J/m² of UV using a Stratalinker 1800 (Stratagene, La Jolla, CA) and incubated at 26° along with the YPD-MMS plate and an untreated YPD plate that served as a control. In addition, one YPD plate was incubated at 37°. After incubation for 3 days, the growth and color of the colonies on the 37°, UV, and MMS plates were compared to the control plates, and the colonies showing a growth defect or reddish color were recovered as candidate mutants and streaked for single colonies on YPD or YPAD plates containing erythrosine B. A colony from each candidate mutant was retested using the semiquantitative assays described below.

Semiquantitative assays:
Cells from a single colony were streaked on three YPD plates. One plate was incubated at 26° and served as a control. One plate was irradiated with 50 J/m² of UV and incubated at 26°. And the third plate was incubated at 37°. Cells from the same colony were also picked into sterile water and fourfold serial dilutions were prepared. Five-microliter aliquots of each dilution were spotted onto three YPD plates, one of which contained 0.01% MMS and another of which was irradiated with 50 J/m² of UV. These three plates were incubated at 26°. After incubation for 2 days, the plating efficiency and extent of the growth on each plate compared to the control plates were classified as follows: for UV and MMS sensitivity: s, 10- to 10²-fold decrease in plating efficiency; ±, decreased <10-fold. For temperature-sensitive growth: ss, no or little growth; s, very small colonies (<10% the size of the colonies on the control plates); ±, smaller colonies (10 to 50% the size of the colonies on the control plates).

Quantitative UV survival tests:
Overnight cultures were inoculated into liquid YPAD and grown to early logarithmic phase (2–5 x 10⁶ cells/ml). The cells were harvested by centrifugation, suspended in sterile water, recentrifuged, and resuspended in sterile water at 1 x 10⁶ cells/ml, followed by brief sonication to disrupt aggregates. Serial dilutions were prepared in sterile water, aliquots were spread on YPD plates, and the plates were UV-irradiated as indicated. The plates were then incubated at 26° in the dark for 4 days before counting colonies.

Quantitative MMS survival tests:
Cells were prepared as described under quantitative UV survival tests to obtain a cell suspension containing 1 x 10⁶ cells/ml. MMS was added to 0.5 ml of cell suspension at the indicated concentrations followed by incubation for 40 min at room temperature. An equal volume of freshly prepared 10% (wt/vol) NaS₂O₃ was added, and serial dilutions of the cells were prepared in sterile water and spread on YPD plates. Colonies were counted after incubation at 26° for 4 days.

Flow cytometry:
Cells were grown in liquid YPAD to early logarithmic phase (2–5 x 10⁶ cells/ml) at 26°, and then the culture was divided into two portions: one portion was incubated at 26° and the other portion was incubated at 37°. Cells were withdrawn at each indicated time point and prepared for fluorescence-activated cell sorting (FACS) analysis as previously described ( JOHNSON and KOLODNER 1995 ). The resulting samples were analyzed using a FACScan flow cytometer with the CellFIT cell cycle analysis program (version 2.01.2, Becton Dickinson, Franklin Lakes, NJ) by the Core Flow Cytometry Facility, Dana-Farber Cancer Institute. A portion of each sample also was examined by light microscopy.

DNA sequencing:
The entire region corresponding to the RFA1 open reading frame in mutant pKU1 plasmids was sequenced with Taq DNA polymerase and dye terminators using an ABI 373 DNA sequencer (Applied Biosystems, Foster City, CA) by Molecular Biology Core Facility, Dana-Farber Cancer Institute. Six sequencing primers were used and were synthesized by the Molecular Biology Core Facility: 5'-GGCGAAACCAGCAAGAAGAC, 5'-CTCAGAGCATCCAAATGAAACC, 5'-GAAGCCAAAGTATACTATGTATC, 5'-TAAAGGTGTTCGTGTGACGGA, and 5'-ATTTTGCATATCCTGCCTGTTC were complementary to the transcribed strand of RFA1 at nucleotides -39 to -20, 372 to -393, 751 to -773, 1131 to -1151 and 1701 to -1722, respectively, and 5'-CTCATATGTTACATAGATTAAATAG was complementary to the nontranscribed strand of RFA1 at nucleotides 1886 to 1911 (numbering is from the first base of the initiation codon, which is 1) (HEYER et al. 1990 ).

Physical analysis of mating type switching and single-strand annealing:
Strains W303-1A and an isogenic derivative containing the UV ^S, MMS^S rfa1-t11 allele were transformed with pJH727 containing GAL1::HO CEN4 LEU2. Galactose inductions were carried out as described previously (CONNOLLY et al. 1988 ). Briefly, the strains were grown to 10⁷ cells/ml in yeast extract peptone-lactate medium. At time zero and subsequent time points, DNA was extracted from aliquots using a glass bead protocol (CONNOLLY et al. 1988 ). The strains were induced by the addition of galactose (2%, w/v) for 30 min, followed by the addition of glucose (2% w/v) to down-regulate the expression of GAL::HO. At 0 hr and 5 hr aliquots of cells were diluted in sterile H₂O and plated on YPD. Colonies were later scored for their phenotypes. DNA samples were digested with StyI, electrophoresed through agarose gels, and analyzed by Southern blotting using a MAT-distal probe (pJH364) (WHITE and HABER 1990 ). A Molecular Dynamics (Sunnyvale, CA) PhosphorImager was utilized to measure band intensities. Single-strand annealing was assayed using the rfa1-t11 derivative of tNS1373M essentially as described above for the physical analysis of mating type switching except for the following procedural modifications. Glucose was not added to down-regulate the expression of GAL::HO. The DNA was digested with EcoRI, blotted, and probed with the Bsu36-SacI fragment from lacZ to detect the plasmid products of HO cleavage and single-strand annealing. Colonies were grown on YPD and replica-plated onto media lacking leucine to assay for plasmid loss after induction of HO endonuclease.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of rfa1 mutants:
We have attempted to systematically analyze the in vivo functions of RPA in DNA metabolism by screening for rfa1 mutants that confer ts, UV ^S, or MMS^S phenotypes. Ts mutants were selected to identify mutations causing defects in DNA replication. UV ^S mutants were selected to identify mutations causing defects in nucleotide excision repair and/or recombinational repair. And MMS^S mutants were selected to identify mutations that cause defects in recombination as well as possibly defects in other types of DNA repair (ESPOSITO and WAGSTAFF 1981 ; PETES et al. 1991 ). The mutants isolated in these screens have subsequently been tested for mutator phenotypes, and the results of these studies will be described elsewhere.

The yeast strain RKY2102 was cotransformed with RFA1 fragments that had been amplified by PCR under mutagenic conditions and pRS415 derivative vector fragments, resulting in reconstruction of an RPA1 plasmid by in vivo homologous recombination. We then screened 15,000 Leu⁺ 5-FOA-resistant transformants for ts, UV^S, and MMS^S phenotypes, first using erythrosine B staining as a growth defect indicator (the first screening) and second using semiquantitative measurements of the expected phenotypes. Fifty-four candidate mutants were obtained for further study, some of which had more than one phenotype. No mutants were obtained in control experiments in which RFA1 was not amplified by PCR prior to the transformation step.

The resulting candidate mutant plasmids were retested by isolating plasmid DNA from each mutant and transforming it back into RKY2102 followed by plasmid shuffling. Nineteen out of 22 of the UV ^S and/or MMS^S mutant plasmids conferred a similar degree of sensitivity to that of the original mutants from which the plasmid had been isolated. The sensitivity of all 19 mutants was suppressed in the presence of pRPA1 carrying the wild-type RFA1 gene, indicating that the mutants were all recessive. After plasmid rescue, only 5 out of 10 ts mutant plasmids tested conferred a ts phenotype. All 5 of these ts mutants were recessive to wild type. In total, 24 mutants were obtained and 21 were studied further; specific examples of the analysis of these mutants are presented below, and a summary of the properties of all of the mutants, including quantitative data, are presented in Figure 8 at the end of this article.

View larger version (23K): [in this window] [in a new window] [Download PPT slide]   Figure 1. —Survival of representative mutants treated with either UV-irradiation or MMS. Quantitative UV and MMS survival tests were performed as described in materials and methods. The number shown next to each symbol indicates the mutant allele tested. The wild-type RFA1 control strain, RKY1900, carries the RFA1 gene on pKU1 and has the same rfa1 deletion mutation on the chromosome as the rfa1 mutant strains tested; all of the rfa1 alleles examined were present on a plasmid in this experiment.

View larger version (56K): [in this window] [in a new window] [Download PPT slide]   Figure 2. —Sensitivity of UVS and MMSS mutants to HO endonuclease. The indicated mutant strains transformed with the GAL-HO plasmid pJH132 were grown in SC medium lacking glucose, uracil, and leucine and containing 3% glycerol and 2% sodium lactate (pH 5.5) to early logarithmic phase. Approximately 7 x 104 cells from each culture were spotted in duplicate onto an SC plate lacking glucose, uracil, and leucine and containing 2% galactose to induce the GAL-HO gene (left panel) or were spotted onto an SC plate lacking uracil and leucine as control (right panel). The plates were then incubated at 26° for 4 days. As in Figure 1, RKY1900 was used as a wild-type control strain.

View larger version (38K): [in this window] [in a new window] [Download PPT slide]   Figure 3. —RFA1 is required for mating type switching after induction of HO endonuclease. An rfa1-t11 mutant containing the rfa1 allele at the chromosomal locus and a wild-type strain were induced to switch mating type by expression of a GAL::HO fusion. DNA was extracted at the time points shown, digested with StyI, electrophoresed under neutral conditions, blotted and hybridized with a MAT-specific probe (hatched box). At 0 hr the probe hybridizes to the MATa fragment (0.93 kb) and the MATdistal fragment (4.3 kb). After GAL::HO expression a smaller HO-cleaved fragment (0.72 kb) appears, followed at later time points by the appearance of a MATa product band (1.88 kb). This band is larger because it lacks a StyI site in the Ya sequence that is present in Ya. In the rfa1 mutant the product band is faint but detectable.

View larger version (33K): [in this window] [in a new window] [Download PPT slide]   Figure 4. —RFA1 is required for efficient single-strand annealing. (A) The plasmid pNSU208 (15 kb) contains two 0.24-kb direct repeats composed of lacZ sequences with an HO cut site in between. Cleavage by HO endonuclease is followed by 5' to 3' degradation of DNA leaving a 3' single-stranded tail. The complementary lacZ sequences can anneal and be processed to yield a deletion product. (B) Single-strand annealing was initiated in vivo when a double-strand break was created by induction of the GAL::HO endonuclease gene in an rfa1-t11 mutant containing the rfa1 allele at the chromosomal locus and a wild-type strain. Cleavage of the 9.4-kb pNSU208 EcoRI fragment by HO endonuclease resulted in the production of 6.2- and 3.2-kb fragments by 0.5 hr and yielded a 7.8-kb product fragment in the wild-type strain. The amount of product formed was greatly reduced in the rfa1-t11 mutant.

View larger version (45K): [in this window] [in a new window] [Download PPT slide]   Figure 5. —Some rfa1 mutants display a strong temperature-sensitive phenotype. The indicated mutant and wild-type strains were streaked on two YPD plates that were incubated at 26° (left panel) and at 37° (right panel) for 3 days, respectively. The allele number is indicated next to each streak. When the allele designation is followed by -c, it indicates the allele at the chromosomal locus, whereas -p indicates the allele carried on an ARS CEN plasmid. For wild-type control strains, RKY 1900 was used as wt-p and W303-1A was used as wt-c.

View larger version (39K): [in this window] [in a new window] [Download PPT slide]   Figure 6. —FACS analysis of DNA content. Cells were grown in liquid YPAD to early logarithmic phase at 26°, and then the culture was divided into two portions: one portion was incubated at 26° and the other portion was incubated at 37°. Cells were withdrawn after 4 hr, stained with propidium iodide, and analyzed using a FACScan flow cytometer. The results obtained with the rfa1-t33 allele integrated at the chromosomal locus and the wild-type strain W303-1A are shown here. Identical results were obtained when similar analysis was performed with the rfa1-t6 allele at the chromosomal location.

View larger version (31K): [in this window] [in a new window] [Download PPT slide]   Figure 7. —Morphological analysis of wild-type and rfa1-t33 cells at the restrictive temperature. A portion of the culture used in the experiments shown here was withdrawn at each indicated time point and examined by light microscopy. Cells in which the bud size was less than one-third the size of the mother cell were classified as "small buds," while the cells harboring larger buds were classified as "large buds." Cells having multiple buds or protruded buds were classified as "other." The results obtained with the rfa1-t33 allele integrated at the chromosomal locus and the wild-type strain W303-1A are shown here. Identical results were obtained when similar analysis was performed with the rfa1-t6 allele at the chromosomal location.

View larger version (39K): [in this window] [in a new window] [Download PPT slide]   Figure 8. —Summary of the phenotypes caused by rfa1 mutations and illustration of the positions of the mutations relative to the positions of known RFA1 structural features. The coordinates of the different functional regions are taken from the papers discussed in the introduction and discussion (ERDILE et al. 1991 ; ADACHI and LAEMMLI 1992 ; BROWN et al. 1994 ; GOMES and WOLD 1995 , GOMES and WOLD 1996 ; ISHIAI et al. 1996 ; KIM et al. 1996 ; LIN et al. 1996 ; PARK et al. 1996 ; PHILIPOVA et al. 1996 ; BOCHKAREV et al. 1997 ). The positions of the indicated amino acid changes are given relative to the structural features of the protein and are based on the following mutational data: Ile14Ser, T41G; Phe15Leu, T43C; Met49Thr, T146C; Lys45Glu, A133G; Asp121Gly, A362G; Arg109Cys, C325T; Phe124Ser, T371C; Glu562Val, A1685T; Ser373Pro, T1117C; Gly207Arg, G619A; Tyr255Cys, A764G; Leu221Pro, T662C; Phe222Ser, T665C; His274Arg, A821G; Ser230Pro, T688C; Ser351Pro, T1051C; Leu602Ser, T1805C; Ser391Pro, T1171C; Leu410Ser, T1229C; Ser569Cys, A1705T; Lys493Glu, A1477G; Lys494Glu, A1480G; Phe537Ile, T1609A. The mutational changes are annotated in the following manner: open triangle, residue mutated is not conserved among S. cerevisiae, human, Drosophila, and Xenopus RPA1; closed triangle, residue mutated is conserved among S. cerevisiae, human, Drosophila, and Xenopus RPA1. In addition, the following silent nucleotide changes were found: rfa1-t22, A255T; rfa1-t69, T195C; rfa1-t23, A630G; rfa1-t63, C1074G; rfa1-t155, T1533C. UV- and MMS-sensitivity were determined at a fixed dose of either 75 J/m2 of UV-irradiation or 40 min of incubation in 0.5% MMS as described in materials and methods and in Figure 1: the values given are the fold reduction in survival relative to the wild-type strain. (The fraction of wild-type cells that survived UV treatment was 0.21 and the fraction of wild-type cells that survived MMS treatment was 0.34.) HO sensitivity summarizes the data presented in Figure 2: —, average of >20 colonies per patch survive upon HO induction (wild-type level); s, average of 10–20 colonies per patch survive upon HO induction; ss, average of 6–10 colonies per patch survive upon HO induction; sss, average of 0–5 colonies per patch survive upon HO induction; nt, not tested. Temperature sensitivity refers to whether a consistent temperature-sensitive phenotype like that illustrated in Figure 5 was seen as defined in materials and methods under Semiquantitative assays. The doubling times presented were determined by growing the indicated strains in liquid YPD medium at 26° and monitoring cell growth both by counting cell numbers and by plating onto YPD plates to determine the number of viable cells present. All of the data summarized in this figure were generated with strains in which the rfa1 alleles were present on a plasmid.

UV ^S and MMS^S mutants:
Nineteen mutants showing increased sensitivity to UV-irradiation and/or MMS were identified and analyzed in greater detail using quantitative tests. Figure 1 shows the survival curves obtained after treatment of representative mutants with a range of doses of either UV-irradiation or MMS, and the survival of each of these mutants after treatment with a fixed dose of either UV-irradiation or MMS is presented in Figure 8. All experiments were performed using rfa1 mutant strains (RKY 2102 derivatives) in which the mutant rfa1 allele of interest was present in the strain on an ARS CEN plasmid. In addition, control experiments were performed using RKY 2102 and RKY 1900 derivatives carrying the wild-type RFA1 gene on an ARS CEN plasmid. Selected mutations were also transferred to the chromosomal RFA1 locus and similarly analyzed to ensure that the plasmid-born alleles and chromosomal alleles behaved similarly.

All of the mutants obtained were sensitive to both UV-irradiation and MMS, even though the initial qualitative tests suggested the existence of mutants that were sensitive to killing by only one of the two agents. The mutants showed a wide range of sensitivity to both UV-irradiation and MMS at the doses tested; survival of mutants relative to the wild-type strain ranged from 10^-3 to 0.5 for UV-irradiation and from 10^-4 to 0.7 for MMS (Figure 8). There was a direct correlation between the degree of sensitivity to UV-irradiation and to MMS for each mutant. An interesting feature of the data is that the weakly and moderately UV ^S alleles showed similar sensitivity to MMS, whereas the strongly UV ^S alleles showed an even greater sensitivity to MMS; this is more apparent when the data presented in Figure 8 are reranked in order of increasing sensitivity.

The experiments discussed above (Figure 1 and Figure 8) were performed with strains carrying the rfa1 allele on an ARS CEN vector. Although we obtained the same results when either YPAD or Leu drop-out media that select for the marker on the plasmid (LEU2) was used, loss of the plasmid could conceivably have affected the results. To eliminate this possibility, the rfa1-t11 allele was integrated at the chromosome RFA1 locus, and the UV-sensitivity and MMS-sensitivity of the resulting strain were determined. Survival of this strain to killing by UV-irradiation and MMS was 3.85 x 10^-4 and 1.02 x 10^-4, respectively, compared to the wild-type parental strain, which was similar to that obtained with rfa1-t11 allele on an ARS CEN plasmid at the same doses of UV-irradiation and MMS (Figure 8). Similar results were obtained with the rfa1-t49 allele as well as three additional alleles that will be described elsewhere.

Some of the mutant strains, especially the strongly UV ^S and MMS^S mutants, had an increased doubling time compared to the wild-type strains tested (Figure 8). This could be due to failure to repair endogenous DNA damage in these mutants resulting in triggering of a DNA damage checkpoint. Alternately, these mutant strains might also have defects in DNA replication that cause reduced growth rates. To examine these possibilities, we analyzed early logarithmic phase cells of strains carrying either the rfa1-t11 chromosomal allele or the rfa1-t11 plasmid allele using both FACS and light microscopy. Compared to the wild-type control strain (RKY 2102), both mutant strains showed no significant increase in the proportion of cells present in a specific phase of the cell cycle such as S phase or G2 phase (data not shown). This suggests that the rfa1-t11 allele does not cause a defect in DNA replication or trigger a DNA damage checkpoint.

Sensitivity of UVS and MMSS mutants to HO endonuclease:
We anticipated that mutants that were sensitive to killing by both UV-irradiation and MMS might also have defects in genetic recombination because both UV-irradiation and MMS are known to produce damage in DNA that can be repaired by genetic recombination (PETES et al. 1991 ). To examine whether the UV ^S and MMS^S mutants had such defects, we induced double-strand breaks at MAT by galactose induction of HO-endonuclease and determined whether the mutants were killed by this treatment; the mutants showing very weak UV ^S and MMS^S phenotypes were not tested. Two different types of tests were performed: one measured the plating efficiency of the mutants on plates containing galactose and the other measured the effect of transient induction of HO on survival. These tests allowed the identification of three classes of mutants.

The transformation efficiency of mutants carrying the rfa1-t21, rfa1-t155, and rfa1-t18 alleles with pJH132 (pGAL-HO) was much lower than with the other strains examined when the transformants were selected on plates containing glucose. The pJH132-transformed cells had growth defects even on glucose plates without galactose and were unable to survive on galactose plates (Figure 2). This suggests that these mutants were extremely sensitive to double-strand breaks and that either the extremely low level of HO expression in the presence of glucose is sufficient to kill some of the cells or that HO expression is increased in these rfa1 mutants.

Strains carrying the rfa1-t11, rfa1-t22, rfa1-t48, and rfa1-t69 alleles and the pGAL-HO plasmid showed clear differences in growth in the presence and the absence of galactose (Figure 2). These strains grew somewhat more slowly than a wild-type strain on glucose plates, and the survival of these mutant strains on galactose plates was much lower than the wild-type control strain. The strain carrying the rfa1-t11 allele was examined in greater detail by exposing the strain to galactose for different periods of time and then plating the cells onto glucose plates to measure survival. Ninety percent of the cells carrying the rfa1-t11 allele were killed after 4 hr of exposure to galactose compared to the wild-type control strain, confirming the extreme sensitivity of this mutant to even transient induction of double-strand DNA breaks at MAT (data not shown).

To further investigate the role of RPA, we monitored recombination induced by a double-strand break by Southern analysis. Mating type switching was induced in mutant and wild-type cultures by galactose induction of HO endonuclease from pJH727. After induction, DNA samples were extracted and analyzed by Southern blotting. Figure 3 shows efficient cutting at MATa by the HO endonuclease followed by the appearance of a product band (MAT) in the wild-type control. The chromosomal rfa1-t11 mutant generates a comparable level of HO-cleaved DNA but much less product DNA. Over time the HO-cleaved DNA in the rfa1-t11 mutant disappeared, consistent with the cleaved DNA being degraded and the cells dying as was observed on transient exposure to galactose (discussed above). The MAT product band produced in the rfa1-t11 mutant represents 4.4% of the total MATa and sequences, compared with 56% for the wild-type strain. The kinetics of appearance of MAT product in the rfa1-t11 mutant strain are indistinguishable from the wild-type strain. This defect in switching in the rfa1-t11 mutant was confirmed by examination of the colonies produced after 5 hr of HO-induction. In the rfa1-t11 mutant only 7% of the colonies maintaining the GAL::HO plasmid successfully switched from MATa to MAT (55% switched in the wild-type strain). The rfa1-t11 allele is not a null allele and retains some types of RFA1 activity. Nevertheless, both the Southern blot analysis and the colony assays indicate that the rfa1-t11 mutant is highly defective in its ability to carry out double-strand break-induced recombination.

Induction of HO endonuclease had a less clear effect on strains carrying the rfa1-t49, rfa1-m51, rfa1-t23, and rfa1-t83 alleles (Figure 2), consistent with the observation that these alleles were among the moderate UV ^S and MMS^S alleles identified. This suggests that these mutations only cause minor defects in homologous recombination, resulting in only moderate sensitivity to UV-irradiation, MMS, and HO endonuclease. Alternately, these mutations may not affect homologous recombination and may only cause defects in repair pathways like nucleotide excision repair or base excision repair.

Defects in single-strand annealing in the rfa1-t11 mutant:
Single-strand annealing (SSA) is a process in which single-stranded DNA regions are generated on each side of a double-strand break in the form of 3' tails (SUGAWARA and HABER 1992 ). When complementary sequences on each side of the break become single stranded, they can anneal and subsequently be processed to result in a single copy, while deleting the intervening sequence. Given RFA1's involvement in mating type switching, a process mediated by double-strand break repair, we tested whether RFA1 has a role in SSA by utilizing an HO cut site situated between direct repeats of a 0.24-kb E. coli lacZ sequence carried on a centromeric plasmid. A Southern hybridization analysis of DNA extracted from a culture undergoing SSA revealed that the rfa1-t11 mutant was impaired in its ability to carry out SSA relative to a wild-type strain (Figure 4). Although a small amount of product could be formed, the mutant was 8.5 times less efficient than wild type in carrying out SSA based on a densitometric analysis of the blots. This defect was also manifested by the loss of HO endonuclease-cleaved plasmid substrate in the mutant strain. The number of colonies retaining the plasmid after induction compared to those before induction was 0.97 for wild type but only 0.12 for the rfa1-t11 mutant.

Analysis of ts rfa1 mutants:
We identified 5 ts mutants (of 10 mutants that were retested) with which it was possible to demonstrate that the ts phenotype was associated with an rfa1 mutation. Two of these mutant alleles (rfa1-t6 and rfa1-t33) were successfully transferred to the chromosomal RFA1 locus, whereas it was not possible to transfer the other three mutant alleles to the chromosomal RFA1 locus (Figure 5). As shown in Figure 6, strains carrying the rfa1-t33 and rfa1-t6 alleles at either the chromosomal locus or on an ARS CEN plasmid had a clear temperature-sensitive phenotype. Strains carrying the rfa1-t33 allele, both when integrated on the chromosome or on the ARS CEN plasmid, grew normally at the permissive temperature, while at the restrictive temperature they grew only at heavily streaked regions and formed no visible individual colonies. Strains carrying the chromosomal rfa1-t6 allele grew more slowly at the permissive temperature than strains carrying this allele on an ARS CEN plasmid. Strains carrying the chromosomal rfa1-t6 allele did not grow at all at the restrictive temperature, whereas strains carrying this allele on an ARS CEN plasmid grew only at heavily streaked regions and formed tiny colonies at the restrictive temperature. Thus, the ts phenotype caused by the integrated alleles (particularly in the case of the rfa1-t6 allele) was more severe than the plasmid-born alleles. This suggests that the reason we were not able to transfer three of the ts rfa1 alleles to the chromosomal locus was that they were lethal when integrated at the RFA1 chromosomal locus.

The ts mutants did not initially appear to be sensitive to killing by UV-irradiation or MMS when they were characterized using quantitative assays as plasmid-born alleles. However, when the rfa1-t33 and rfa1-t6 alleles were transferred to the chromosomal locus, the resulting strains became sensitive to UV-irradiation and MMS. When these alleles were present on an ARS CEN plasmid, they conferred only limited sensitivity to UV-irradiation and MMS: Survival of a strain carrying a plasmid-born rfa1-t33 allele was 0.143 and 0.116 after treatment with 75 J/m² of UV-irradiation and 0.5% MMS, respectively, and survival of a strain carrying a plasmid-born rfa1-t6 allele was 8.27 x 10^-2 and 8.4 x 10^-2, respectively, which was similar to that observed for the wild-type control strain (Figure 8). However, when these alleles were transferred to the chromosomal RFA1 locus, they caused significantly increased sensitivity to both UV-irradiation and MMS at the standard doses tested: Survival of the rfa1-t33 strain after treatment with 75 J/m² of UV-irradiation and 0.5% MMS was 1.01 x 10^-2 and 1.83 x 10^-3, respectively, and survival of the rfa1-t6 strain was 3.48 x 10^-3 and 2.05 x 10^-3, respectively. These results suggest that expression of RPA1 from the chromosome is somewhat lower than that on an ARS CEN plasmid and that repair and/or recombination is more sensitive to RPA1 levels than DNA replication in the case of the ts alleles rfa1-t6 and rfa1-t33, as compared to the repair-defective alleles rfa1-t11 and rfa1-t49 where no difference between plasmid and chromosomal alleles was observed (see above).

To characterize the ts phenotype conferred by the rfa1-t33 and rfa1-t6 alleles in more detail, we grew cultures of strains carrying these alleles at the permissive temperature, shifted logarithmically growing cultures to the nonpermissive temperature, and examined the resulting changes in cellular morphology and DNA content (Figure 6 and Figure 7). The experiments presented were performed with a wild-type control strain and a strain carrying the rfa1-t33 allele at the chromosomal locus; however, identical results were obtained with both the rfa1-t33 and rfa1-t6 alleles, regardless of whether they were on an ARS CEN plasmid or at the chromosomal RFA1 locus (data not shown).

Figure 6 shows FACS analysis of the DNA content of wild-type and rfa1-t33 cells (identical data were obtained with the rfa1-t6 allele but are not presented due to space considerations). After 4 hr at the restrictive temperature, the rfa1-t33 strain showed a clear difference in the distribution of cells having differing DNA contents compared to the wild-type strain. The rfa1-t33 strain showed a significant decrease in the proportion of cells having either a 1N or 2N DNA content and a significant increase of the proportion of cells having greater than 1N but less than 2N DNA content compared to the wild-type control strain. This effect was seen to a lesser extent after 2 hr of incubation at the nonpermissive temperature and did not increase with continued incubation beyond 4 hr (data not shown). These results suggest that most of the rfa1-t33 cells did not complete DNA replication and accumulated prior to M phase at the restrictive temperature. This was confirmed by analyzing the cells by light microscopy (Figure 7, discussed below); identical data were obtained with the rfa1-t6 allele but are not presented due to space considerations. After 4 hr at the restrictive temperature, 70% of the rfa1-t33 cells had arrested with a single large bud. After staining with DAPI, these large-budded cells were observed to have a single nucleus at the neck of the bud (data not shown). The large-budded cells appeared to have been derived from the small-budded cells presented in the culture at the time of temperature shift because the proportion of small-budded cells decreased by the same amount that the proportion of large-budded cells increased, whereas the proportion of nonbudded cells present in the culture did not appear to change. In addition, approximately 3% of the arrested rfa1-t33 cells had protruded buds or dumbbell-shaped buds like those observed after sporulation of rfa1 null mutants (HEYER et al. 1990 ). Control experiments with the wild-type parental strain showed a normal distribution of cells at different stages of the cell cycle at both temperatures. From these results, we conclude that the ts rfa1 mutants rfa1-t33 and rfa1-t6 have defects in the elongation stage of DNA replication at the restrictive temperature, resulting in cell cycle arrest at a premitotic stage, or they arrest prior to M phase at a step that causes a block in DNA replication.

In the experiments presented in Figure 6 and Figure 7, we observed that prior to shifting to the restrictive temperature, 15–20% of the rfa1-t33 cells had 1N DNA content and a similar proportion of the cells did not have buds consistent with 15–20% of the cells in G1. After 2 and 4 hr at the restrictive temperature, the proportion of such rfa1-t33 cells did not change. This suggests that either these cells are dead or that they are unable to enter S phase and initiate DNA replication. Similar results were obtained with rfa1-t6 cells (data not shown).

DNA sequence analysis of rfa1 mutations:
To determine the nature of the rfa1 mutations causing the observed altered phenotypes, we sequenced the rfa1 gene of all 19 UV ^S and MMS^S alleles and both ts alleles. Figure 8 and Figure 9 summarize the mutations identified and the predicted amino acid changes they cause. Every allele had at least one mutation in the RFA1 gene. Thirteen alleles had either a single nucleotide change or two changes of which one was a silent change, allowing clear identification of the causal mutation. Eight alleles had two potentially significant nucleotide changes. There were no identical alleles; however, rfa1-t155 could easily be related to rfa1-t21 by the subsequent accumulation of a second, silent nucleotide change.

View larger version (45K): [in this window] [in a new window] [Download PPT slide]   Figure 9. —Alignment of different RPA sequences and amino acid changes caused by rfa1 mutations. Alignment of the predicted RFA1 amino acid sequences for the ssDNA binding domains 1 and 2 of S. cerevisiae, human, Drosophila and Xenopus RFA1 (see PHILIPOVA et al. 1996 ; BOCHKAREV et al. 1997 ). Identities are indicated by boxes, and the positions of the amino acid changes from Figure 8 are indicated by circles and are labeled with the relevant allele numbers. S. cerevisiae and human are indicated by y and h, respectively. Domains 1 and 2 are indicated as RPA1A and RPA1B, respectively. Down-arrows indicate ssDNA-binding contacts in domain 1 and up-arrows indicate ssDNA-binding contacts in domain 2.

Twenty-three mutations were identified, and most of them changed amino acids that were conserved among yeast, human, Drosophila melanogaster, and Xenopus lavis (see Figure 8 for an analysis of the mutations, including a list of mutations affecting conserved amino acids). It is possible to ascribe significance to some of these mutations based on what is known about the structure of RFA1. Five mutations mapped in the region necessary for pol stimulation by RFA (KIM et al. 1996 ). Three mutations causing very weak phenotypes mapped in the region between the regions required for pol stimulation and DNA binding. Ten mutations, including two causing a high degree of sensitivity to UV-irradiation and MMS (rfa1-t48 and rfa1-m51), mapped in either domain 1 or domain 2 of the conserved region of RFA1 required for ssDNA binding (HEYER et al. 1990 ; GOMES and WOLD 1995 ; PHILIPOVA et al. 1996 ; BOCHKAREV et al. 1997 ). The mutations rfa1-t21, rfa1-t155, and rfa1-t141 altered amino acids in the putative zinc finger of RFA1, which is not necessary for DNA-binding activity (GOMES and WOLD 1995 ; LIN et al. 1996 ) but could act to mediate protein-protein interactions (GALCHEVA-GARGOVA et al. 1996 ; RODGERS et al. 1996 ; SHEPARD et al. 1996 ; SZABO et al. 1996 ). Finally, at least one mutation mapped in the C-terminal region required for RPA holocomplex formation. Understanding the exact effects of each mutation on the function of RPA1 will require a detailed analysis of the biochemical properties of the mutant RPA1 proteins. However, these data do allow us to make some predictions about the functions of the different regions of RPA1, and these will be discussed below.

In the case of alleles having two nucleotide changes, it is not clear which nucleotide change is the causal mutation. However, some insights into the nature of these alleles can be obtained by comparing the properties of these alleles. The rfa1-t22 allele had two nucleotide changes, one of which in position 43 was in common with the rfa1-t47 allele that had only that change. Because the rfa1-t47 allele causes a lower degree of UV- and MMS-sensitivity than the rfa1-t22 allele, it seems likely that the second nucleotide change present in the rfa1-t22 allele (position 146) also contributes to the mutant phenotype caused by the rfa1-t22 allele. The rfa1-t11 and rfa1-t69 alleles cause a similar degree of UV- and MMS-sensitivity. This suggests that the common nucleotide change at position 133 might be the causal mutation, while the second nucleotide change in the rfa1-t69 allele at position 362 may not contribute to the mutant phenotype. One of the strong ts alleles, rfa1-t6, shared the same nucleotide change at position 1051 with the rfa1-t83 allele, which had an additional change at position 1805. Because the rfa1-t83 allele caused moderate UV- and MMS-sensitivity in addition to a ts phenotype, the latter change at position 1805 might be responsible for its UV- and MMS-sensitive phenotypes, while the former change at position 1805 might be the causal mutation for temperature sensitivity. It will be necessary, however, to use site-directed mutagenesis to construct appropriate single mutants to fully understand the nature of the alleles containing more than one potentially significant nucleotide change.

The rfa1-t33 and rfa1-t92 alleles had one nucleotide change in common, a T to C change at position 1117 that changed the serine to a proline. The rfa1-t33 allele caused a strong ts phenotype, whereas the rfa1-t92 allele just showed slight UV-sensitivity (Figure 8). These results suggest that the second mutation in rfa1-t92 suppressed the ts phenotype caused by the serine to proline change and caused the weak UV-sensitive phenotype, possibly in combination with the serine to proline change.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In an effort to understand the role of RPA in DNA metabolism and to begin to identify regions of RPA required for different RPA-dependent processes, we have used a plasmid shuffle method to isolate random mutations in the RFA1 gene encoding the 70 kDa subunit of RPA. These mutants were then screened for ts, UV ^S, and MMS^S, and 24 mutants having a variety of different phenotypes were obtained. In addition, we have retained a large number of other uncharacterized rfa1 mutants that will be useful for future studies. Previous studies described the first rfa1 mutants and some of the phenotypes caused by rfa1 mutations; however, these studies identified only a limited number of rfa1 mutants (HEYER et al. 1990 ; BRILL and STILLMAN 1991 ; LONGHESE et al. 1994 ; FIRMENICH et al. 1995 ; SMITH and ROTHSTEIN 1995 ). The preliminary analysis of the rfa1 mutants described here has documented a variety of phenotypes caused by rfa1 mutations, including several causing much stronger replication and recombination defects than previously described. Based on the results obtained thus far, it is likely that we will be able to use the mutant collection described here in conjunction with considerable additional experimentation to define in greater detail different roles of RPA in DNA metabolism.

Biochemical studies of RPA have shown that it stimulates the homologous pairing activity of several different homologous pairing proteins, SEP1 (HEYER et al. 1990 ; ALANI et al. 1992 ), HPP1 (MOORE et al. 1991 ), and RAD51 (SUNG 1994 , SUNG 1997 ; SUGIYAMA et al. 1997 ) and interacts with proteins required for excision repair (HE et al. 1995 ; LEE et al. 1995 ; LI et al. 1995 ; MATSUDA et al. 1995 ; MATSUNAGA et al. 1996 ) and genetic recombination (FIRMENICH et al. 1995 ; PARK et al. 1996 ). Consistent with this, we were able to isolate a number of rfa1 alleles that conferred different degrees of sensitivity to UV-irradiation and MMS (Figure 1 and Figure 8). Interestingly, most of these alleles were not temperature-sensitive and did not appear to cause defects in DNA replication. Similarly, many of the mutations that caused replication defects had only limited effects on repair of UV- and MMS-induced damage to DNA. These results suggest that there are a number of ways in which mutations in RFA1 may alter the properties of RPA without changing its ability to function in DNA replication. This may reflect the possibility that the DNA and protein-protein interaction requirements for RPA1 may differ significantly between DNA replication and DNA recombination/repair. For example, the ability of RPA1 to interact with and stimulate the activity of XPA, XPF, XPG, and RAD52 (FIRMENICH et al. 1995 ; HE et al. 1995 ; LEE et al. 1995 ; LI et al. 1995 ; MATSUDA et al. 1995 ; MATSUNAGA et al. 1996 ; PARK et al. 1996 ) likely reflects a function that is required for recombination and repair but not for replication. An interesting parallel in this regard is the E. coli SSB, where the ssb-113 mutation alters its DNA-binding properties so as to cause repair and recombination defects but not replication defects (MEYER and LAINE 1990 ).

Previous genetic studies have provided evidence that RPA functions in genetic recombination (LONGHESE et al. 1994 ; FIRMENICH et al. 1995 ; SMITH and ROTHSTEIN 1995 ) and interacts with at least one recombination protein. Two of these studies reported the isolation of rfa1 alleles that caused modest, less-than-10-fold reductions in direct repeat recombination (SMITH and ROTHSTEIN 1995 ) and mitotic intrachromosomal recombination (LONGHESE et al. 1994 ), respectively. One of these alleles also stimulated recombination between direct repeats in a rad1 rad52 double mutant (SMITH and ROTHSTEIN 1995 ). A third study reported the isolation of an rfa1 allele (rfa1-44) that caused a strong defect in a double-strand break-induced plasmid by chromosome gene conversion (FIRMENICH et al. 1995 ). Subsequent studies have shown that the rfa1-44 allele caused ~50% failure in mating type switching (J. E. HABER and N. SUGAWARA, unpublished data), a result that is not surprising because the mutant isolation procedure used involved induction of HO endonuclease in a MAT strain that should be lethal in the presence of mutations causing strong defects in double-strand break repair. We found that many of the UV ^S and MMS^S mutants described here showed some degree of sensitivity to killing by HO endonuclease compared to the wild-type control strain (Figure 2), suggesting they might be highly defective in recombinational repair of double-strand breaks. More detailed analysis of one UV ^S and MMS^S mutant (rfa1-t11) using genetic and biophysical assays that measure mating type switching showed that this allele caused over a 90% decrease in mating type switching in physical assays (Figure 3), consistent with a strong defect in double-strand break repair. The rfa1-t11 allele also caused ~90% decrease in SSA recombination; this is the first report of such an effect of an RFA allele. Interestingly, the rfa1-t11 allele was not the most HO endonuclease-sensitive, UV ^S, or MMS^S allele we obtained (Figure 2 and Figure 8), suggesting that there are a number of mutants among our collection that are even more recombination defective than rfa1-t11. More detailed analysis of these mutants will be performed in the future to more completely determine their effects on different types of recombination.

Many of the rfa1 mutants we isolated confer a UV ^S phenotype. This is consistent with the view that RPA might play a direct role in the excision repair of UV damage. Such a role for RPA is based on the observations that human RPA is required for excision repair in vitro (COVERLEY et al. 1991 ) and that RPA1 interacts with proteins required for excision repair (ABOUSSEKHRA et al. 1995 ; HE et al. 1995 ; LEE et al. 1995 ; LI et al. 1995 ; MATSUDA et al. 1995 ; MATSUNAGA et al. 1996 ). However, the UV ^S phenotypes we observed could be due to defects in either excision repair or recombinational repair of UV damage. Given the MMS-sensitivity and HO-endonuclease sensitivity of many of the UV ^S mutants we obtained, it would not be surprising if the UV ^S phenotype observed reflected a defect in recombinational repair rather than excision repair. Additional analysis will be required to distinguish between these possibilities.

RPA was originally identified because of its requirement for in vitro DNA replication, suggesting that it is an essential DNA replication factor (WOBBE et al. 1987