Assessing the requirements for nucleotide excision repair proteins of Saccharomyces cerevisiae in an in vitro system.

Nucleotide excision repair (NER) is the primary mechanism by which both Saccharomyces cerevisiae and human cells remove the DNA lesions caused by ultraviolet light and other mutagens. This complex process involves the coordinated actions of more than 20 polypeptides. To facilitate biochemical studies of NER in yeast, we have established a simple protocol for preparing whole cell extracts which perform NER in vitro. As expected, this assay of in vitro repair was dependent on the products of RAD genes such as RAD14, RAD4, and RAD2. Interestingly, it was also dependent upon proteins encoded by the RAD7, RAD16, and RAD23 genes whose precise roles in NER are uncertain, but not the RAD26 gene whose product is believed to participate in coupling NER to transcription. Replication protein A (RPA/Rpa), known to be required for NER in human cell extracts, was also shown by antibody inhibition and immunodepletion experiments to be required for NER in our yeast cell extracts. Moreover, yeast cells with temperature-sensitive mutations in the RFA2 gene, which encodes the 34-kDa subunit of Rpa, had increased sensitivity to UV and yielded extracts defective in NER in vitro. These data indicate that Rpa is an essential component of the NER machinery in S. cerevisiae as it is in mammalian cells.

Nucleotide excision repair (NER) 1 is a versatile DNA repair strategy found ubiquitously in prokaryotes and eukaryotes. NER is capable of removing a broad spectrum of DNA lesions caused by physical and chemical mutagens (1)(2)(3)(4)(5). Failure to remove DNA lesions from the genome as a result of defective NER may lead to cancer-susceptibility, as exemplified by the hereditary human disease xeroderma pigmentosum (XP) (2)(3)(4)(5). NER involves the concerted actions of several different enzymatic activities and can be arbitrarily divided into the following steps: damage recognition, incision and excision of the lesion and its flanking DNA, and repair DNA synthesis to fill in the resulting single-stranded gap. Many of the proteins encoded by XP genes and their evolutionarily conserved RAD gene homologs in Saccharomyces cerevisiae are now known to function in the early steps of excision repair, participating in the removal of DNA lesions prior to the repair DNA synthesis step. Some of the proteins involved in the repair DNA synthesis step of excision repair also function in the replication of cellular DNA and include proteins such as proliferating cell nuclear antigen (6 -8) and replication protein A (9 -14).
Both yeast and mammalian replication protein A (Rpa/RPA) are trimeric complexes consisting of polypeptides of approximately 70, 34, and 14 kDa (15)(16)(17)(18)(19). RPA, a single-stranded DNA-binding protein, might function in the DNA synthesis step of NER as it does in cellular and viral DNA replication. Recent studies (11-13, 20 -22) have suggested, however, that RPA participates in both the early (damage recognition, incision, excision) steps as well as the late (repair synthesis) step of NER. The human RPA complex is now known to interact directly with XPA (20 -23), the human homolog of the yeast damage-recognition protein Rad14 (24,25). This interaction enhances the ability of the RPA-XPA complex to bind to damaged DNA (20,21). Following the damage-recognition step, incisions flanking the lesion are made by two endonuclease activities, in humans XPG (26,27) and the ERCC1-XPF complex (28 -30), and in yeast Rad2 (31) and Rad1-Rad10 (32,33). Two helicase activities encoded by the XPD and XPB genes (34,35) in humans and RAD3 and RAD25 in S. cerevisiae (36 -39) are also involved in the incision/excision steps prior to repair synthesis. In addition to proliferating cell nuclear antigen and RPA, repair DNA synthesis also involves the participation of replication factor C (11,14), polymerase ␦ or ⑀ (11,14,40), and one of the DNA ligases (11,14).
Biochemical analyses of NER have depended largely on the availability of cell-free extracts that can support excision repair. The in vitro system developed by Wood and his colleagues (41) has proven to be instrumental in the dissection of the NER pathway in human cells. On the other hand, understanding excision repair in S. cerevisiae has relied mainly on the genetic analyses of rad mutants. The potential for biochemical analysis of NER in yeast cells has not yet been fully realized owing in part to the lack of a simple in vitro repair system. The only existing cell-free system from S. cerevisiae, that described by Wang et al. (42,43), involves the preparation of separate nuclear and whole cell extracts. Hence, it seemed desirable to develop a simple in vitro preparation from yeast cells capable of NER and applicable for use with a variety of different yeast strains. Here we report the development of such a system using a whole cell extract prepared from S. cerevisiae. We have utilized this in vitro system to demonstrate a requirement for the products of the genes RAD7, RAD16, and RAD23. Furthermore, we have also exploited this in vitro repair system to assess the role of yeast Rpa in NER. In addition, yeast strains with point mutations in the RFA2 gene were constructed and characterized. Taken together, our biochemical and genetic results indicate that, as in human cells, Rpa plays an important role in the process of NER in S. cerevisiae cells.

EXPERIMENTAL PROCEDURES
Yeast Whole Cell Extracts-The S. cerevisiae strains used in this study were BJ2168 (from Dr. J. Segall, University of Toronto), LP2899, a gift of Dr. L. Prakash, and rad mutant strains, MGSC131 (rad4⌬::URA3), MGSC139 (rad14⌬::LEU2), MGSC104 (rad7⌬::LEU2), W303236 (rad16⌬::URA3), MGSC101 (rad23⌬::URA3), MGSC102 (rad26⌬::HIS3) and their isogenic parental strain W303-1B (44) (kindly provided by Dr. J. Brouwer, Leiden University), and EMY75 (rad2⌬::URA3 from L. Prakash). The procedure for the preparation of NER-proficient yeast extract was a modification of protocols originally intended for transcription studies (45). Yeast cultures were grown at 27°C in complete medium (YEPD: 1% yeast extract, 2% Bacto-Peptone, 2% glucose) with vigorous shaking. Cells were harvested at an OD 600 of 2 by first chilling in ice water and then centrifuging at 4000 rpm for 4 min in a Sorval H-6000A rotor. The cells were then washed once in ice-cold water and once in extraction buffer (0.2 M Tris, pH 7.5, 0.39 M (NH 4 ) 2 SO 4 , 10 mM MgSO 4 , 20% (v/v) glycerol, 1 mM EDTA, 1 mM dithiothreitol) containing various protease inhibitors (phenymethylsulfonyl fluoride, 1 mM; benzamidine hydrochloride, 2 mM; pepstatin A, 3.5 g/ml; leupeptin, 1 g/ml; bestatin, 0.35 g/ml; and aprotinin, 10 g/ ml). The drained cell pellet was scraped into a syringe, then extruded directly into liquid nitrogen and stored at Ϫ70°C. Frozen cells were broken by manual grinding under liquid nitrogen using a ceramic mortar and pestle. Grinding continued until the material was reduced to powder. After grinding, the frozen powder of broken cells was mixed with 1 volume of cold extraction buffer supplemented with protease inhibitors and allowed to thaw at 4°C. The cell lysate was then centrifuged at 120,000 ϫ g for 2 h at 4°C. The clear supernatant was recovered and (NH 4 ) 2 SO 4 added to 2.94 M by the addition of 337 mg of solid (NH 4 ) 2 SO 4 /ml of lysate over the course of 1 h. The suspension was stirred for another 30 min, and the precipitated protein pelleted by centrifugation at 40,000 ϫ g for 15 min. The pellet was resuspended in a small volume (approximately 50 l/g of cells) of dialysis buffer (20 mM HEPES pH 7.5, 20% (v/v) glycerol, 10 mM MgSO 4 , 10 mM EGTA, 5 mM dithiothreitol) plus protease inhibitors. The sample was then dialyzed against the same buffer plus 1 mM phenymethylsulfonyl fluoride for 12-16 h. The dialysate was centrifuged to remove precipitated protein and the resulting supernatant was collected and stored at Ϫ70°C until use. Protein concentrations were determined by the Bio-Rad colorimetric assay using bovine serum albumin as standard.
In Vitro Assay of Nucleotide Excision Repair-DNA from plasmids pUC18 (2.7 kilobase pairs) and pGEM-3Zf(ϩ) (3.2 kilobase pairs) was isolated by alkaline lysis and CsCl-ethidium bromide equilibrium centrifugation. The pUC18 DNA was treated with N-acetoxy-2-acetylaminofluorene (AAAF) and repurified on a 5-20% sucrose gradient as described by Wang et al. (42). Reaction mixtures (50 l) contained 300 ng of AAAF-treated pUC18 and 300 ng of control pGEM-3Zf(ϩ) DNA, 45 mM HEPES-KOH (pH 7. Ci of [␣-32 P]dCTP (3000 Ci/mmol), 2 mM ATP, 40 mM disodium phosphocreatine, 2.5 g of creatine kinase, 3.4% glycerol, 18 g of bovine serum albumin, and 250 g of protein as yeast whole cell extract (typically 6 -8 l). Reactions were incubated at 28°C for 2 h. Plasmid DNA was purified from the reaction mixtures, linearized by digestion with HindIII, analyzed on a 1% agarose gel, and autoradiographed as described by Wang et al. (42). To quantitate the extent of repair synthesis, dried gels were exposed to either storage Phosphor screens or autoradiography films. The amount of radioactivity in each band was quantitated by phosphoimaging analysis or densitometry and comparison to known radioactivity standards. For the antibody inhibition experiments shown in Fig. 4, the indicated amount of preimmune or anti-yeast Rpa antiserum was preincubated with the cell extracts for 15 min at 28°C before addition of reaction buffer and plasmid DNAs.
Yeast Rpa Antibody and Immunodepletion-The polyclonal antibodies against yeast Rpa used in these studies were raised in a rabbit by injection of recombinant yeast Rpa protein. The depletion of yeast Rpa from whole cell extracts was effected using an immunodepletion procedure described by Adachi and Laemmli (46). Briefly, phosphate-buffered saline-washed Protein A-Sepharose CL-4B matrix was incubated with equal amounts of preimmune or anti-yeast Rpa serum at room temperature for 2 h. The matrices were washed and incubated with 4 volumes of yeast cell extract in the presence of an ATP-regenerating system (46) at 4°C for 1 h. The Sepharose was then briefly pelleted in an Eppendorf tube and the supernatant was collected as the depleted extract. As estimated by a Western blot, more than 90% of the yeast Rpa protein in the extract was removed (data not shown).
Expression of Recombinant Yeast Rpa-Plasmids pJM124, pJM223, and pJM329 contain, respectively, the RFA1 gene, RFA2 cDNA, and the RFA3 gene under the control of the phage T7 promoter of plasmid pET11a (47) and are described in Ref. 19. The T7 promoter, gene, and transcription terminator of these plasmids can be moved as a BamHI/ BglII cassette. The 1.2-kilobase pair BamHI/BglII cassette of pJM223 was inserted into the BamHI site of pJM329 to create pJM332. The resulting 1.9-kilobase pair BamHI/BglII cassette of pJM332 was then inserted into the BglII site of pJM124 to create pJM126 containing the three RFA genes on a single plasmid, each with its own T7 promoter. pJM126 was transformed into BL21(DE3) for protein expression. BL21(DE3) cells containing pJM126 were grown in LB containing 0.1 mg/ml ampicillin at 22-37°C to an OD of 0.5-0.8. Isopropyl-1-thio-␤-D-galactopyranoside was then added to 0.4 mM and the induction continued for 3 h. Cells were harvested and the bacterial pellet resuspended in one-tenth volume of A buffer (18) with 50 mM NaCl and 1 mg/ml lysozyme. Following incubation at 4°C for 15 min, Nonidet P-40 was added to a final concentration of 0.1%. The lysate was subjected to a total of 3 freeze-thaw cycles using dry ice/ethanol and swirling in a 37°C bath. Chromosomal DNA was dispersed by sonication (3 ϫ 1 min) in volumes of approximately 20 ml on ice. The extract was clarified by centrifugation at 20,000 ϫ g for 20 min at 4°C. The extract produced from 1 liter of culture was loaded on a 100-ml phosphocellulose column equilibrated in buffer A with 50 mM NaCl and the column washed with 300 ml of buffer A with 50 mM NaCl. Escherichia coli SSB is found in the flow-through of this column. Rpa was eluted with a 600-ml gradient from 50 to 800 mM NaCl in buffer A and assayed for unwinding activity as described previously (48). Rpa elutes at approximately 200 mM NaCl from this column. Active fractions were pooled, NaCl added to 500 mM, and the sample loaded onto a 5-ml ssDNA-cellulose column equilibrated in buffer A with 500 mM NaCl. This column was washed with 15 ml of buffer A with 500 mM NaCl, 25 ml of buffer A with 750 mM NaCl, and 25 ml of buffer A with 1.5 M NaCl and 50% ethylene glycol. The protein peak from this last step contains approximately 1 mg of recombinant Rpa from a 1-liter culture. Approximately 50 g of recombinant Rpa was then loaded onto a 5-ml 15-35% glycerol gradient in buffer A containing 100 mM NaCl. A parallel gradient contained a similar amount of Rpa purified from yeast. The gradients were centrifuged in an SW50.1 rotor at 49,000 rpm for 21 h. Fractions (130 l) were collected from both gradients, 8-l aliquots assayed for unwinding activity and 14-l aliquots subjected to SDS-PAGE followed by silver staining. The unwinding assay was performed essentially as described by Tsurimoto et al. (48). A final reaction volume of 20 l containing 30 mM HEPES-KOH (pH 8.0), 7 mM MgCl 2 , 0.5 mM dithiothreitol, 4 mM ATP, 40 mM creatine phosphate, 0.4 g of creatine phosphokinase, 0.4 g of bovine serum albumin, 0.08 g of calf thymus topoisomerase I, 0.3 g of pSV011 plasmid DNA, 0.5 g of SV40 T antigen, and the indicated amounts of Rpa was assembled on ice and then placed at 37°C for 15 min. The reaction was stopped by placing it on ice. Four l of a solution of 4% SDS, 40 mM EDTA, and 6 mg/ml Pronase was added and the tubes placed at 37°C for 30 min. The reaction products were then directly resolved by electrophoresis in a 0.8% agarose gel, stained with ethidium bromide, and photographed. For experiments employing recombinant Rpa in the in vitro assays of nucleotide excision repair, recombinant Rpa protein was purified through an Affi-Gel Blue column, an hydroxylapatite column, and an ssDNA-cellulose column as described for the purification of human RPA by Henricksen et al. (49).
Isolation of rfa2 Mutants-Details concerning the identification and characterization of rfa2 mutants will be presented elsewhere. 2 Briefly, the entire RFA2 gene on plasmid pJM215 (19) was subjected to 35 cycles of amplification with Taq DNA polymerase in the presence of 1 mM MnCl 2 and the universal forward and reverse sequencing primers. The polymerase chain reaction product was digested with BamHI and SalI and ligated into the yeast centromeric plasmid pRS415. This library of mutagenized RFA2 plasmid DNA was amplified in bacteria and introduced into strain SBY105 which carries a deletion of the RFA2 gene but is kept alive by plasmid pJM218 (19) carrying the RFA2 cloned into the yeast centromeric plasmid YCp50. Transformants were replicaplated to plates containing the drug 5-fluoroorotic acid and placed at 37°C and 25°C. Strains showing no growth after 2 days at 37°C were identified and the RFA2 plasmid rescued from the corresponding 25°C colony by transformation into E. coli. Mutant plasmid DNA was then reintroduced into strain SBY105 (19) to confirm that the temperaturesensitive growth defect was plasmid-dependent. Strains passing this test were tested for loss of viability in a liquid culture at 37°C. Typically these ts strains showed 10 -20% viability following 4 h exposure to 37°C.
Analysis of UV Sensitivity-Cells were grown at 25°C overnight in YPD medium, diluted in sterile water, and plated in duplicate on YPD plates and immediately irradiated with a Stratalinker UV cross-linker (Stratagene). The plates were then incubated at 25°C for 3 days and surviving colonies were counted.

Nucleotide Excision Repair in Yeast Cell Extracts-
The availability of human cell-free extracts capable of performing nucleotide excision repair (41) has permitted major advances in our understanding of this pathway of DNA repair. The genetic analysis of the sensitivity to UV of S. cerevisiae has also provided an excellent approach for studying NER in eukaryotes (1,2). However, the full potential of using S. cerevisiae and its many mutant rad strains in studies of NER has not been fully exploited. Biochemical analyses of NER in yeast may have been hampered by the lack of a simple method of preparing yeast extracts capable of repair in vitro. As the human extract system was developed originally for studying in vitro transcription, we tested whether yeast extracts capable of in vitro transcription (45) would also perform nucleotide excision repair.
In this protocol, yeast cells were quickly frozen and then ground in liquid nitrogen. After addition of minimal amounts of buffer, the thawed extract was then clarified by a centrifugation step, further concentrated by ammonium sulfate precipitation, solubilized, and dialyzed (see "Experimental Procedures"). Nucleotide excision repair in the yeast cell extracts was monitored by the incorporation of radiolabeled nucleotides into plasmid DNA during repair DNA synthesis. DNA damage was introduced into plasmid substrates by treatment with AAAF which forms DNA adducts known to be corrected by the excision repair pathway (50). An untreated plasmid was also included in each reaction to monitor repair-independent nucleotide incorporation. As shown in Fig. 1A, damage-dependent repair synthesis was detected in extracts made from three commonly used laboratory S. cerevisiae strains. The extent of damage-dependent repair synthesis increased with the amount of protein added. An amount of extract corresponding to 250 g of protein in an assay volume of 50 l produced good signals above the background level of incorporation of radiolabeled nucleotide into the untreated plasmid and was therefore routinely used in subsequent experiments. Using these assay conditions, extracts from wild-type cells routinely incorporated 130 -190 fmol of dCMP into the AAAF-treated plasmid DNA during the 2-h incubation period. Background incorporation into the untreated control plasmid by these wild-type extracts was usually in the range of 20 -40 fmol. This in vitro NER activity, like that reported by Wang et al. (42), was absolutely dependent on the addition of ATP (Fig. 1C, lanes 3-5). Omission of added Mg 2ϩ ions in the assay also reduced incorporation significantly (Fig. 1C, lanes 1 and 2). The residual incorporation (lane 2) likely reflects the contribution of Mg 2ϩ in the buffer used to prepare the cell extract.
We first established that the preferential incorporation of radiolabeled deoxynucleoside triphosphate into AAAF-treated DNA seen in the assays of Fig. 1A reflected bona fide nucleotide excision repair. To do this, we examined the dependence of our in vitro system on the products of RAD genes known to be involved in the excision repair pathway. The preferential incorporation of radioactivity into the AAAF-treated DNA was reduced to near background levels in extracts prepared from rad14, rad4, and rad2 deletion strains (Fig. 1B, lanes 2, 3, and   7). Furthermore, when extracts prepared from rad4 and rad14, or rad2 and rad4 mutant cells were mixed in different ratios, the defects in excision repair were corrected (Fig. 1B, lanes 4 -6  and 7-9). This strict dependence of our cell-free system on known repair gene products including one of the incision endonucleases and the restoration of repair synthesis in the mixed extracts established that our simple preparation of S. cerevisiae FIG. 1. In vitro NER in yeast whole cell extracts. The NER activity of yeast cell extracts was assayed by detecting DNA repair synthesis. The incorporation of radioactively labeled deoxynucleoside triphosphate into control or AAAF-treated DNA isolated from repair reaction mixtures was detected by autoradiography after agarose gel electrophoresis. In the upper panels the DNAs were visualized by ethidium bromide staining; the lower panels show the autoradiograms of the gels. A, NER activity of extracts made from three different wild-type strains. The amount of protein used in 50-l reactions is indicated. The femtomoles of dCMP incorporated into AAAF-treated DNAs in lanes 1-9 were 60, 146, 144, 42, 137, 160, 53, 179, and 187, respectively. B, NER activity in the extracts from the indicated isogenic wild-type and rad mutants as well as in mixtures of different rad mutant extracts. The femtomoles of dCMP incorporated into AAAFtreated DNAs in lanes 1-9 were 169, 15, 4, 171, 88, 106, 165, 31, and 170, respectively. C, dependence of DNA repair synthesis during NER on Mg 2ϩ (lanes 1 and 2), ATP (lanes 3 and 4), and an ATP-regenerating system (lanes 3 and 5). The indicated components were omitted from the reaction mixtures in lanes 2-5. The femtomoles of dCMP incorporated into AAAF-treated DNAs in lanes 1-5 were 149, 61, 3, 52, and 61, respectively. extract is indeed capable of in vitro nucleotide excision repair.
We also assessed whether our yeast cell-free repair system was dependent upon the products of the yeast RAD7, RAD16, and RAD23 genes. The functions of the proteins encoded by these three RAD genes remain largely unknown, although RAD7 and RAD16 have been shown to be necessary for repair of the non-transcribed strand of transcribed genes and for repair of transcriptionally silent regions of the genome (44,51). Rad23 may promote the formation of a repair complex (52). Fig.  2A shows that extracts made from the rad7 and rad16 deletion strains were completely defective for NER in our cell-free system while the rad23 deletion strain had a reduced ability to repair damaged DNA. In contrast to these results with rad7, rad16, and rad23 strains, an extract prepared from a rad26 deletion strain retained its ability to perform NER in vitro, an expected result given the evidence that the Rad26 protein functions exclusively in the subpathway of transcription-coupled repair (44,53). To confirm that the deficient in vitro NER in the rad7 and rad16 extracts was not due simply to poor extract preparation from these particular strains, in vitro complementation experiments were also performed. When extracts from rad4 and rad7 or rad16 cells were mixed, the defects in excision repair were corrected (Fig. 2B, lanes 6 and  7). Curiously, however, when rad7 and rad16 extracts were mixed, in vitro NER was not restored (lane 5). Because the Rad7 and Rad16 proteins appear to function together in the repair of silent regions of the genome, extracts prepared from cells singly deleted for either function may also have acquired a deficiency in the other protein.
Overexpression and Purification of Recombinant Yeast RPA-We and others have shown previously that human RPA functions at early steps in NER, and not exclusively at the DNA synthesis step of repair. In this study, we wished to address the role of yeast Rpa in nucleotide excision repair using the cellfree system described above. In order to facilitate these studies, we first expressed and purified recombinant yeast Rpa. The three subunits of yeast Rpa were co-expressed in E. coli using a T7 RNA polymerase-based expression system and the assembled trimeric Rpa complex was purified in four chromatographic steps to near homogeneity. To verify that recombinant yeast Rpa functioned like authentic yeast Rpa we characterized the purified protein after glycerol gradient sedimentation. Fractions from the gradient were assayed for the ability to cooperate with SV40 T-antigen to unwind plasmid DNA containing the SV40 origin of DNA replication. SDS-PAGE analysis of the glycerol gradient fractions revealed co-sedimentation of the three subunits indicating that the proteins exist in a complex (Fig. 3A). No free subunits were detected in this preparation and the sedimentation velocity of the complex was identical to authentic yeast Rpa run in a parallel gradient (data not shown). These fractions of recombinant Rpa supported the SV40 unwinding assay (Fig. 3B) as originally described for Rpa purified from yeast cells (18,48). In addition, we also demonstrated that the purified recombinant yeast Rpa was able to bind single-stranded DNA in a mobility-shift assay (data not shown).
Requirement for Yeast Rpa in NER-To explore the role of yeast Rpa in excision repair, polyclonal antibodies raised against recombinant yeast Rpa were tested for their effects on DNA repair in our cell-free extract system. In Western blots these rabbit antibodies recognized the 70-kDa Rpa1 and the FIG. 2. In vitro NER activity in extracts prepared from RAD7,  RAD16, RAD23, and RAD26 mutant strains. A, NER activity, detected by incorporation of radiolabeled deoxynucleoside triphosphate into DNA during repair synthesis, was assessed using extracts made from isogenic wild-type, rad7, rad16, rad23, and rad26 deletion strains of S. cerevisiae. The femtomoles of dCMP incorporated into AAAFtreated DNAs in lanes 1-5 were 154, 14, 24, 49, and 172, respectively. B, complementation of defective NER activity of rad7 and rad16 mutant extracts by the addition of rad4 extract, and lack of complementation when rad7 and rad16 extracts were mixed in a 1:1 ratio. The femtomoles of dCMP incorporated into AAAF-treated DNAs in lanes 1-7 were 130, 24, 28, 22, 28, 57, and 134, respectively. Top panels, DNA stained with ethidium bromide; bottom panels, the autoradiograms of the agarose gels. 34-kDa Rpa2 subunits of the trimeric Rpa efficiently, but bound to the 14-kDa Rpa3 subunit poorly (data not shown). Preincubation of yeast extracts with preimmune serum had little effect on their in vitro repair activity (Fig. 4A, lanes 2-4). However, addition of equal amounts of the rabbit anti-Rpa serum caused a marked inhibition of repair incorporation (lanes 5-7). Furthermore, when the extracts were preincubated with purified recombinant yeast Rpa prior to antisera addition, the inhibitory effects of the anti-Rpa antibodies were overcome (Fig. 4B, compare lanes 6 and 7 to lane 5). Addition of exogenous Rpa had minimal effect on the repair activity in the presence of preimmune serum (Fig. 4B, lanes 2-4).
To provide further evidence for the involvement of yeast Rpa in excision repair, we also used the anti-Rpa antibodies to deplete endogenous Rpa from our yeast extracts and then assayed for repair activity. As monitored by Western blotting (data not shown), more than 90% of the yeast Rpa was depleted from the extract by anti-Rpa antibodies immobilized on protein A beads, but not by control protein A beads preincubated with preimmune serum. The extracts depleted for Rpa failed to support efficient NER (Fig. 5, lane 2), whereas a mock-depleted extract efficiently performed NER (lane 1). Addition of recombinant yeast Rpa restored repair activity to the Rpa-depleted extract (compare lanes 4 and 6 to lane 2). Collectively, these data show that Rpa is required for nucleotide excision repair in yeast extracts as was the case for human RPA in human cell extracts (11-13, 20 -22).
Defective NER in rfa2 Mutants-To provide in vivo evidence for the involvement of yeast Rpa in nucleotide excision repair, we assessed the DNA repair phenotype of yeast strains with conditional-lethal alleles of the RFA2 gene, encoding the 34-kDa subunit of yeast Rpa. The RFA2 gene was mutagenized using a polymerase chain reaction-based strategy and transformed back into yeast. Following removal of the wild-type copy of RFA2, clones conferring a temperature-sensitive phenotype were identified (details to be presented elsewhere). 2 Two alleles, rfa2-214-15 and rfa2-210-3, lost viability quickly at 37°C and were chosen for further analysis. Stable mutant strains were constructed by integrating a single copy of these alleles into the yeast genome as the sole source of the 34-kDa subunit in these cells. The DNA repair capacities of these two rfa2 mutant strains were examined, first by assessing their sensitivity to UV irradiation when grown at the permissive temperature. The UV dose-survival curves revealed that both of the rfa2 mutant strains showed increased sensitivity to UV irradiation, and at doses of 50 -150 J/m 2 , they exhibited a 10 -20-fold decrease in viability compared to wild-type (Fig. 6). Neither strain was as sensitive to UV irradiation as the rad4 mutant strain included for comparison. This heightened sensitivity to UV irradiation of these two rfa2 mutant strains provided additional genetic evidence for the involvement of Rpa in the repair of UV-induced DNA damage.
To characterize excision repair in these rfa2 mutant strains, we prepared cell-free extracts from each of the strains grown at 25°C and assessed them for their ability to perform in vitro DNA repair. Extracts from both rfa2 mutant strains were deficient in our assay of excision repair (Fig. 7A, compare lanes 2 and 3 to the isogenic parental control in lane 1). As expected, addition of recombinant yeast RPA to a defective extract restored in vitro repair activity (Fig. 7B).

DISCUSSION
In this study, we have modified a protocol originally developed for assay of in vitro transcription by RNA polymerases I, II, and III (45) to make an NER-proficient yeast extract. Since these procedures appear to maintain the functional integrity of cellular components, this protocol may have overcome problems with the apparent lability of certain components such as the Rad2 protein which characterized the dual extract in vitro repair system described by Wang et al. (42). We have demonstrated that our yeast extracts preferentially incorporate radiolabeled nucleotides into damaged DNA. This assay appears to detect bona fide excision repair activity since extracts made from mutant strains such as rad14, rad4, and rad2, known to be defective in vivo in yeast NER (13,24), failed to preferentially incorporate nucleotides into the damaged DNA substrate. As these extracts support in vitro transcription by RNA polymerase II only with the addition of ribonucleotide triphosphates (data not shown), and do not require the RAD26 gene product for NER, the signals detected in this study are not likely due to transcription-coupled repair.
This simple system using an easily prepared whole cell extract from a single strain of yeast may prove as useful in biochemical studies of NER as has the analogous mammalian whole cell extract. Although reconstitution of the initial incision/excision steps of NER with purified S. cerevisiae components has recently been reported (13), that system does not support the DNA synthesis step of NER. Our whole cell extract system supports the complete NER reaction. It may also facilitate detailed studies of the roles in excision repair of proteins such as the RAD7, RAD16, RAD23, and RAD26 gene products.
Using this system we have demonstrated that extracts made from rad7, rad16, and rad23 strains failed to efficiently repair AAAF-induced DNA damage. In contrast, an extract made from a rad26 deletion strain retained wild-type levels of in vitro excision repair, in keeping with the demonstrated role of RAD26, the homolog of the human Cockayne syndrome group B gene, in transcription-coupled repair (44,54). Since our yeast transcription extracts also permit efficient transcription by RNA polymerase II, it may be possible, by using a DNA substrate containing an RNA polymerase II promoter, to employ this system to study the role of the Rad26 protein in transcription-coupled repair.
The in vitro repair deficiency of the rad23 extract is also consistent with the notion that the Rad23 protein might function as an assembly factor for a repair complex, mediating the association between Rad14 and TFIIH (52). When the incision step of NER was reconstituted with purified components, there was a requirement of Rad23 protein and/or the tightly associ-ated Rad4 polypeptide (13). This reconstituted assay of incision, however, did not appear to require the Rad7 and Rad16 proteins. The RAD7 and RAD16 gene products have been shown to be absolutely essential for removal of UV-induced pyrimidine dimers from transcriptionally silent regions of the genome and from the nontranscribed strand of transcriptionally active genes in yeast (44,51). These two proteins also contribute to the repair of the transcribed strands of active genes (51). The different requirements for Rad7 and Rad16 proteins in the two in vitro assays, incision or repair synthesis, may reflect the presence of additional factors in our whole cell extract system. One possibility is that, despite the use of the naked DNA substrates in our in vitro assay, this substrate is rapidly assembled into a nucleosomal array, thus necessitating proteins such as Rad7 and Rad16 to relieve a chromatin-dependent repression of NER. A role for the Rad7 and Rad16 proteins in chromatin remodelling in vivo has been suggested (55). The use of our yeast extracts to dissect the different steps of NER may help delineate the precise functions in repair for the Rad7 and Rad16 proteins.
We have used this system to examine the involvement of yeast Rpa in NER. The antibody inhibition and Rpa depletion experiments, taken together with the demonstration of increased UV sensitivity for two mutant rfa2 strains, argue per- suasively for the involvement of Rpa in NER in yeast cells. Other recent reports have suggested that Rpa might function in the damage recognition and incision steps of NER, in addition to any role in the late step of repair DNA synthesis (11-13, 20, 21). That the two rfa2 temperature-sensitive strains used in this study grow at near-normal rates at the permissive temperature of 25°C, and yet appear defective in NER at this permissive temperature both in vivo and in vitro, may indicate that the mutations in the Rpa2 polypeptide do not affect DNA synthesis but rather one of the early steps of NER. A mutation in the yeast RFA1 gene has similarly been reported to confer increased sensitivity to UV without affecting DNA synthesis (56). The yeast Rpa complex may function, as human RPA appears to, at the early step(s) of damage-recognition and incision by making specific and direct contacts with proteins such as Rad14 and Rad2. Further experiments using our in vitro assay of NER and direct tests of protein-protein interactions with these mutant Rpa polypeptides may help account for the rfa2 mutant phenotypes and ultimately lead to an understanding of what appear to be the multiple roles of Rpa in nucleotide excision repair.