The Evolutionarily Conserved Zinc Finger Motif in the Largest Subunit of Human Replication Protein A Is Required for DNA Replication and Mismatch Repair but Not for Nucleotide Excision Repair*

The largest subunit of the replication protein A (RPA) contains an evolutionarily conserved zinc finger motif that lies outside of the domains required for binding to single-stranded DNA or forming the RPA holocomplex. In previous studies, we showed that a point mutation in this motif (RPAm) cannot support SV40 DNA replication. We have now investigated the role of this motif in several steps of DNA replication and in two DNA repair pathways. RPAm associates with T antigen, assists the unwinding of double-stranded DNA at an origin of replication, stimulates DNA polymerases α and δ, and supports the formation of the initial short Okazaki fragments. However, the synthesis of a leading strand and later Okazaki fragments is impaired. In contrast, RPAm can function well during the incision step of nucleotide excision repair and in a full repair synthesis reaction, with either UV-damaged or cisplatin-adducted DNA. Two deletion mutants of the Rpa1 subunit (eliminating amino acids 1–278 or 222–411) were not functional in nucleotide excision repair. We report for the first time that wild type RPA is required for a mismatch repair reaction in vitro. Neither the deletion mutants nor RPAm can support this reaction. Therefore, the zinc finger of the largest subunit of RPA is required for a function that is essential for DNA replication and mismatch repair but not for nucleotide excision repair.

The largest subunit of the replication protein A (RPA) contains an evolutionarily conserved zinc finger motif that lies outside of the domains required for binding to single-stranded DNA or forming the RPA holocomplex. In previous studies, we showed that a point mutation in this motif (RPA m ) cannot support SV40 DNA replication. We have now investigated the role of this motif in several steps of DNA replication and in two DNA repair pathways. RPA m associates with T antigen, assists the unwinding of double-stranded DNA at an origin of replication, stimulates DNA polymerases ␣ and ␦, and supports the formation of the initial short Okazaki fragments. However, the synthesis of a leading strand and later Okazaki fragments is impaired. In contrast, RPA m can function well during the incision step of nucleotide excision repair and in a full repair synthesis reaction, with either UV-damaged or cisplatin-adducted DNA. Two deletion mutants of the Rpa1 subunit (eliminating amino acids 1-278 or 222-411) were not functional in nucleotide excision repair. We report for the first time that wild type RPA is required for a mismatch repair reaction in vitro. Neither the deletion mutants nor RPA m can support this reaction. Therefore, the zinc finger of the largest subunit of RPA is required for a function that is essential for DNA replication and mismatch repair but not for nucleotide excision repair.
Human replication protein A (RPA) 1 is a stable complex of three subunits of Rpa1 (70 kDa), Rpa2 (34 kDa), and Rpa3 (13 kDa). It was first purified from HeLa and 293 cell extracts as an essential component of SV40 DNA replication (1)(2)(3). Binding to single-stranded DNA with high affinity is a hallmark of human replication protein A, and purification of this protein largely takes advantage of this property (3,4).
Rpa1 is the most well characterized subunit among the complex. Rpa1 alone confers the high affinity for single-stranded DNA (5,6), yet only the whole heterotrimeric complex is active in supporting DNA replication (4,7). Rpa1 can be subdivided into 3 domains: an N-terminal domain of protein-protein interaction, a central domain for DNA interaction, and a C-terminal domain for complex formation with Rpa2-3 (8,9).
RPA is highly conserved throughout evolution. Homologous heterotrimeric single-stranded DNA-binding proteins have been identified in nearly all eukaryotes examined (10 -17). All genes reveal significant homology between species at the amino acid level. All of the known Rpa1 homologs contain a conserved putative C4-type zinc finger motif in the C-terminal third of the protein (7,11,18,19). In fact, a point mutation that disrupts the putative zinc finger eliminates DNA replication, confirming the importance of this highly conserved motif (8,20).
RPA is essential for other cellular DNA metabolism involving single-stranded DNA intermediates. This includes DNA repair (21,22) and homologous recombination (10,(23)(24)(25). hRPA's role in DNA replication is largely based on the study of in vitro SV40 DNA replication model. With its interaction with ssDNA, T antigen, and DNA polymerase ␣-primase complex, hRPA assists T antigen to unwind the origin (2, 26 -28) and DNA polymerase ␣-primase complex to synthesize the first Okazaki fragment (29). hRPA is also important in a later elongation step where it stimulates both DNA polymerase ␣ and DNA polymerase ␦ activity and cannot be substituted for by Escherichia coli SSB (30). Besides the interaction with repair proteins (XPA, XPG, and XPF see below) and replication proteins, hRPA has also been implicated in interactions with transcription factors p53 (31,32), Gal4, VP16 (33), RNA polymerase holoenzyme (34), and recombination protein RAD52 (35)(36)(37).
Since DNA repair reactions generally involve DNA synthesis to form a repair patch, it is perhaps not surprising that hRPA takes part in DNA repair and can modulate the repair synthesis step (38). However, the most important function of RPA is at an earlier stage of nucleotide excision repair (21,39), and analysis of nicking of UV-irradiated plasmid DNA during excision repair revealed that RPA is necessary for incision formation (40). RPA is an essential factor for dual incision of damaged DNA with purified human protein components (41,42). The Rpa1 and Rpa2 subunits of hRPA interact with the nucleotide excision repair protein XPA, probably to increase the efficiency of damage recognition, and this would be expected to modulate the excision reaction (22,43,44). In addition, RPA can also modulate the efficiency of cleavage by the incision enzyme XPG and ERCC1-XPF (45). A role for hRPA in general mismatch repair in eukaryotes has not been defined. E. coli SSB protein is required in a reconstituted system of mismatch repair with purified bacterial proteins (46). It therefore seems reasonable that hRPA could be involved in eukaryotic mismatch repair.
To help dissect the role of RPA in DNA replication and repair, we have taken advantage of three defined mutants in the Rpa1 subunit (8). All of the Rpa1 mutants form a stable heterotrimeric complex and show a similar extent of binding to single-stranded DNA under physiological conditions. The present work examines their ability to function during defined steps in DNA replication, nucleotide excision repair, and mismatch repair. The experiments reveal requirements for intact domains in Rpa1 that differ depending on the DNA transaction, suggesting that specific interactions take place during each of the three DNA replication and repair processes examined.

MATERIALS AND METHODS
Protein Purification-The plasmid expressing wild type and various RPA mutants were described in a previous report (8). The proteins were named according to the Rpa1 mutant present in the complex. The proteins were expressed and purified as described (4). The baculovirus expressing DNA polymerase ␣/primase subunits were generous gifts from Drs. Teresa Wang and Ellen Fanning. DNA polymerase ␣/primase was expressed in Hi-5 cells with baculovirus infection and purified by immunoaffinity column chromatography with monoclonal antibody SJK-237-71 specific for the large subunit p180 (47). T Ag was also purified from baculovirus-infected Hi-5 cells by immunoaffinity column chromatography with monoclonal Ab Pab 419 (48). Polymerase ␦ was purified from calf thymus in four steps (DEAE-cellulose, phenyl-Sepharose, S-Sepharose, and Mono-Q chromatography). 2 The conditions in DEAE-cellulose and phenyl-Sepharose columns were as described (49). The active fractions from phenyl-Sepharose eluate were pooled and loaded onto an S-Sepharose column (5 mg/ml bed) pre-equilibrated with buffer D (20 mM potassium phosphate, pH 7.2, 0.5 mM EDTA, 0.1 mM EGTA, 0.5 g/ml leupeptin, 0.2 mM phenylmethylsulfonyl fluoride, 1 mM DTT, 20% glycerol). After the column was washed with 2-3 bed volumes of buffer D containing 0.1 M KCl, DNA polymerase ␦ activity was eluted stepwise with a KCl gradient from 0.1 to 1 M in the same buffer. Active fractions were pooled and dialyzed against buffer E (25 mM Tris-HCl, pH 8.0, 0.025 M NaCl, 1 mM EDTA, 1 mM DTT, 0.1 mM phenylmethylsulfonyl fluoride, 10% glycerol, 0.01% Nonidet P-40, 2 g/ml leupeptin) containing 20% sucrose. The dialyzed enzyme was loaded onto a Mono-Q column pre-equilibrated with buffer E. After washing with 5 bed volumes with buffer E, polymerase ␦ activity was eluted with a NaCl linear gradient from 0.025 to 0.5 M in the same buffer. PCNA-stimulated DNA polymerase activity eluted at 0.2 M NaCl fractions just prior to the protein peak. The resulting polymerase can be stimulated 10-fold by PCNA on a poly(dA)/oligo(dT) template and has no detectable activity on a primed M13 ssDNA template unless PCNA and RF-C were both present. Baculoviruses expressing subunits of human RF-C and calf thymus RF-C protein were the generous gift of V. Podust and E. Fanning (Vanderbilt University) (50).
SV40-based DNA Replication-Replication of a plasmid (pSV010) containing SV40 origin of DNA replication was carried out with 293 cell extracts depleted of RPA (1) and supplemented with various bacterially expressed recombinant RPA holocomplexes as described previously (4). The products were linearized with HindIII and analyzed on a 0.8% alkaline agarose gel. The gel was fixed in 8% (w/v) trichloroacetic acid after electrophoresis and dried. The incorporation of [␣-32 P]dCMP was determined using the DE81 paper and scintillation counting (51).
Pulse Labeling for Early DNA Product-The assay is essentially based on previous reports (52,53). Reaction mixtures (50 l) contained 7 mM MgCl 2 , 0.5 mM DTT, 4 mM ATP, 40 mM creatine phosphate, 30 mM HEPES (pH 7.8), 0.2 units of creatine kinase, 0.6 g of supercoiled SV40 origin-containing plasmid (pSV010), 0.84 g of T Ag, 1.4 g of recombinant RPA preparations as indicated, and RPA-depleted 293 cell lysate. Reaction mixtures were preincubated for 45 min at 37°C in the absence of T Ag to lower T Ag-independent labeling of form II DNA and then further incubated for 15 min after the addition of T Ag. Reaction mixtures were pulse-labeled for 1 min by the addition of 3.5 l of a solution containing [␣-32 P]dCTP (final concentration in the complete reaction, 1 M, 300 cpm/fmol) dATP, dGTP, and dTTP (final concentration of 100 M each) and CTP, GTP, and UTP (final concentration of 200 M each). The reactions were terminated by addition of 30 l of stop solution (60 mM EDTA, 0.3% (w/v) SDS, 2 mg/ml of Pronase) and incubated at 37°C for another 30 min. The products were extracted with phenol/chloroform followed by ethanol precipitation, boiled for 4 min in an equal volume of formamide loading buffer, and analyzed by electrophoresis through 10% polyacrylamide gels containing 8 M urea. The gel was fixed in 10% acetic acid, 10% methanol for 15 min, dried, and analyzed by radiography.
Origin Unwinding Assay-The origin unwinding reaction was carried out in 20-l volumes containing unwinding buffer (30 mM HEPES (pH 8.0), 7 mM MgCl 2 , 4 mM ATP, 0.5 mM DTT, 0.1 mg/ml bovine serum albumin, 40 mM creatine phosphate, and 0.8 units of creatine phosphokinase) in the presence of 72 ng of T Ag, 0.2 g of Sau3A-digested carrier DNA, 50 -60 ng of radiolabeled HindIII-SphI fragment of pSV010, and the indicated amounts of recombinant RPA (54,55). Reaction was carried out at 37°C for 2 h and terminated by adding 10 l of stop solution (described above) followed by continued incubation at 37°C for 30 min. The products were extracted with phenol/chloroform followed by ethanol precipitation and then analyzed by electrophoresis through 6% acrylamide gels.
Enzyme-linked Immunosorbent Assay-The assay was carried out as described previously (56). Briefly, the wells of a 96-well microtiter plate were coated overnight with excess (1 g) recombinant RPA proteins or E. coli single-stranded DNA-binding protein. The wells were then blocked by incubation for 2 h with 3% bovine serum albumin in phosphate-buffered saline. The indicated amounts of purified T Ag were added in 50-l volumes of blocking solution for 2 h. After washing, bound T Ag was detected by Pab 419 monoclonal antibody (diluted 1:1000 in blocking solution), peroxidase-conjugated rabbit anti-mouse antibody (diluted 1:1000 in blocking solution), and the chromogenic substrate 2,2-azino-bis-(3-ethyl-benzothiazoline-6-sulfonic acid) (Sigma).
DNA Polymerase Assays-DNA polymerase ␣ stimulation reactions were carried out in 20-l volumes containing 30 nM M13 ssDNA, 90 nM sequencing primer (U. S. Biochemical Corp.), 20 mM Tris acetate (pH 7.3), 5 mM magnesium acetate, 20 mM potassium acetate, 1 mM DTT, 0.1 mg/ml bovine serum albumin, 1 mM ATP, 0.1 mM dATP, dGTP, TTP, and 0.025 mM dCTP (1000 cpm/pmol). DNA polymerase ␣ was added after a 10-min preincubation period with RPA, and the reaction was continued for another 10 min in 37°C. The incorporation of [␣-32 P]d-CMP was determined by DE81 paper method (51). DNA polymerase ␦ stimulation reactions were carried out in 25-l volumes as described (57). The incorporation of [␣-32 P]dCMP was determined by DE81 paper, and the products were analyzed on a 1% alkaline agarose gel. The gel was processed as mentioned above.

Nucleotide Excision Repair Assays
Repair Synthesis in UV-damaged DNA-The plasmids used were derivatives of pUC vectors, the 3.0-kb pBluescript KS ϩ (Stratagene), and the 3.7-kb pHM14 (58). pBluescript KS ϩ was UV-irradiated (450 J/m 2 ). Both plasmids were treated with E. coli Nth protein, and closed circular DNA was isolated from cesium chloride and sucrose gradients (59). Reaction mixtures (50 l) contained 48 g of human CFII fraction protein depleted of RPA and PCNA (40) mM phosphocreatine (di-Tris salt), 2.5 g of creatine phosphokinase, 3.4% glycerol, 18 g of bovine serum albumin and were incubated at 30°C for 3 h. Plasmid DNA was purified from the reaction mixtures, linearized with BamHI, and loaded on a 1% agarose gel containing 0.3 g/ml ethidium bromide. Data were analyzed by autoradiography with intensifying screens, densitometry, and liquid scintillation counting of the excised bands.
Dual Incision Assay of Cisplatin-damaged DNA-Covalently closed circular DNA containing a single 1,3-intrastrand d(GpTpG)-cisplatin cross-link (Pt-GTG) was prepared as described (60). This DNA substrate was used to analyze the dual incision process of nucleotide excision repair which leads to the excision of characteristically sized platinated oligomers 24 -32 nucleotides in length. Each 150-l reaction mixture contained 144 g of HeLa cell CFII fraction protein, depleted of RPA and PCNA (40), and 4.8 g of the indicated recombinant RPA in buffer containing 45 mM HEPES-KOH (pH 7.8), 70 mM KCl, 7.4 mM 2 T. Tsurimoto, personal communication.
MgCl 2 , 0.9 mM dithiothreitol (DTT), 0.4 mM EDTA, 2 mM ATP, 22 mM phosphocreatine (di-Tris salt), 2.5 g of creatine phosphokinase, 3.4% glycerol, and 18 g of bovine serum albumin. After 5 min at 30°C, either Pt-GTG DNA (750 ng) or control DNA (750 ng) without the cisplatin cross-link was added and incubation continued for 30 min at 30°C. The DNA was purified and digested with XhoI and HindIII for 4 h at 37°C. The reactions were stopped by adding formamide buffer containing bromphenol blue and xylene cyanol. The DNA was denatured at 95°C for 5 min prior to loading on a 12% acrylamide gel and run until the blue dye migrated ϳ30 cm from the wells. DNA was transferred by capillary action for 90 min onto a Hybond-N ϩ membrane soaked in 10 ϫ Tris borate buffer. The membrane was fixed in 0.4 M NaOH for 20 min followed by a 2-min wash in 5 ϫ SSC. The fixed membrane was incubated at 42°C for 16 h in hybridization bottles containing 40 ml of 130 mM potassium phosphate (pH 7.0), 250 mM NaCl, 7% SDS, 10% PEG 8000, and 100 pmol of 32 P-labeled 27-mer oligonucleotide which is complementary to the excised platinated oligomers (60). The membranes were washed for 10 min in 2 ϫ SSC buffer containing 0.1% SDS before exposure of the membrane to x-ray film.

Mismatch Repair Assay
Construction of Phagemid Mutants and Heteroduplex-The mismatch-containing substrates used in these studies were constructed from derivatives of phagemid pBS-SK in which the polylinker between the ApaI and BamHI sites had been replaced by different mutant polylinkers. Heteroduplex substrates were constructed and purified using unpublished procedures similar to those described by others (61). The polylinker regions of the substrates used the following Sequence 1 which contains a GT mispair within overlapping XhoI and NsiI recognition sequences such that repair of the mispair can be monitored by cleavage of the product DNA with either XhoI or NsiI. The substrate contained a single strand break in the NsiI strand at the ScaI site (nucleotide 1106-from the mispair). Importantly, the substrates contain a unique AlwNI site 2.1 kb from the mispair. The construction of the mutant phagemids and the substrate construction methods will be described in greater detail elsewhere.
Mismatch Repair Assay-The repair assay contains 50 g of S100 extract (from 293 cells); 30 mM HEPES (pH 7.8); 7 mM MgCl 2 ; 4 mM ATP, 200 M CTP, GTP, and UTP; 100 M dNTP; 40 mM creatine phosphate; 100 g of creatine phosphokinase/ml; 15 mM sodium phosphate (pH 7.5); and 100 ng of the mismatch substrate (62). The reaction mixture was incubated at 37°C for 2-3 h. After the 2-h incubation, 50 l of stop buffer (0.67% SDS; 0.025 M EDTA) and proteinase K (0.3 mg/ml final concentration) were added. After an additional 15 min incubation at 37°C, the reaction mixture was extracted with phenol/ chloroform and chloroform, and the DNA was precipitated, digested with appropriate enzymes, and run on a 1.2% agarose gel at 65 V for 2.5 h. The fractionated DNA was stained by SYBR green and visualized on a fluoroimager (Molecular Dynamics). The repair results in the restoration of an XhoI or NsiI site, and the repair efficiency observed ranged from 5 to 15%. For antibody neutralization experiments, the extract was incubated with various antibodies for 15 min at 37°C prior to the addition of the mismatch substrate. For RPA rescue experiments, the RPA and the mismatch substrate were added simultaneously.

RESULTS
The DNA Binding Subdomains and the Evolutionarily Conserved Zinc Finger Motif in RPA Are Indispensable for Replication-We have previously described three human RPA complex mutants, m1-616 tRPA (two of the four cysteines in the putative zinc finger of Rpa1 (amino acid 500 and 503) are changed to serine, renamed here as RPA m ), 278 -616 tRPA (deletion of amino acids 1-277 of Rpa1), and ⌬222-411 tRPA (deletion of amino acids 222-411 of Rpa1) (8 To avoid possible batch to batch deviation, different batches of RPA-depleted human cell lysate and recombinant RPA proteins were used in the SV40 DNA replication assay. As previously reported, three RPA mutants consistently showed little replication activity compared with wild type protein 1-616 tRPA when added back to the RPA-depleted cell lysate. Similar results have also been observed by others (20). Replication products of the various RPA mutants were visualized on a 0.8% agarose denaturing alkaline gel after linearization of catenated products with HindIII (Fig. 1). Although the reaction with wild type RPA contained long and short replication products, no products were detected with any of the mutant forms of RPA. Thus both DNA binding subdomains of Rpa1 and the evolutionarily conserved zinc finger are essential for lagging or leading strand DNA synthesis in the replication reaction.
Since DNA replication is the sum of many sub-reactions, the RPA mutants could be functionally active in some but not all of the sub-reactions. Likewise, RPA participates in DNA repair events in addition to DNA replication. Some of the replication defective RPA mutants could still function in DNA repair providing us with tools for comparing the functional requirements of RPA in DNA repair versus DNA replication. We therefore examined the activity of these RPA mutants in DNA replication related sub-reactions and in two types of DNA repair.
T Antigen-RPA Interaction-T antigen (T Ag) is a central molecule of the SV40 DNA replication that binds to the origin of DNA replication and recruits RPA to the origin as it launches the replication process. To study the protein-protein interaction between T Ag and RPA, a modified enzyme-linked immunosorbent assay was employed (56). Increasing amount of T Ag was allowed to interact with a fixed level of immobilized RPA protein or E. coli SSB. The results indicated that all the RPA mutants are able to interact with T Ag (Fig. 2). ⌬222-411 tRPA bound to T Ag to the same extent as wild type RPA, whereas RPA m and 278 -616 tRPA retain about 80 and 50% activity, respectively. Therefore the inability of the mutants to support DNA replication is not explained by a failure to interact with T Ag.
Zinc Finger Motif, but Not the DNA Binding Domains, Is Dispensable for Origin Unwinding-T Ag recognizes the SV40 origin and unwinds the DNA with the assistance from RPA. A radiolabeled SV40 origin-containing duplex linear DNA will be unwound to generate single-stranded DNA products that can be detected as slower migrating species relative to the duplex DNA (Fig. 3A). When increasing amounts of wild type RPA (1-616 tRPA) (lanes 1-4) or zinc finger mutant RPA m (lanes 5-8) were added to the reaction in the presence of T Ag and competitor DNA, the amount of unwound products increased. However, 278 -616 tRPA (lanes 9 -12) showed only low activity, and ⌬222-411 tRPA (lanes [13][14][15][16][17] had no detectable activity (compare with (lane 18, no RPA)). The unwound products were excised from the gel and quantitated by scintillation counting (Fig. 3A). RPA m had at least 80% activity compared with that of wild type RPA. In comparable experiments, both 1-616 tRPA and zinc-finger mutant RPA m unwind up to 50% of the input double-stranded DNA. Thus the zinc finger in RPA is not essential for origin unwinding. E. coli SSB supports T Agmediated origin unwinding (66), but 278 -616 tRPA, which also binds single-stranded DNA under high salt conditions, failed to do so. Thus the salt-resistant DNA binding noted with 278 -616 tRPA is physiologically non-functional, suggesting that both DNA binding subdomains of Rpa1 were essential to bind DNA in a manner that was essential for origin unwinding.
Zinc Finger of Rpa1 Is Dispensable for Stimulating DNA Polymerase ␣ and ␦-RPA is known to physically interact with DNA polymerase ␣-primase complex and stimulate lagging strand DNA synthesis. This stimulatory effect is species-specific, and E. coli SSB does not support this stimulation (30,56). Various levels of RPA proteins and E. coli SSB were added to reactions containing primed M13 ssDNA and DNA polymerase ␣-primase complex (Fig. 4A). Both wild type RPA and the zinc-finger mutant RPA m stimulated DNA synthesis by polymerase ␣ by 8-fold, and higher concentrations of RPA inhibited polymerase ␣ activity completely. This is consistent with a previous report (67). However, neither 278 -616 tRPA nor ⌬222-411 tRPA, which bind to ssDNA with high and low affinity, respectively, stimulated DNA polymerase ␣. Consistent with previous reports, E. coli SSB did not stimulate DNA synthesis by DNA polymerase ␣ (data not shown). E. coli SSB or 278 -616 tRPA, both of which bound to ssDNA with high affinity, did not stimulate DNA polymerase ␣. Thus stimulation of DNA polymerase ␣ requires some features of RPA other than high affinity DNA binding. The N-terminal domain of Rpa1 has been implicated in protein-protein interaction with DNA polymerase ␣. However, ⌬222-411 tRPA contains this domain but still did not stimulate polymerase ␣, indicating that high affinity DNA binding is also required from RPA for this activity. In contrast, RPA m which contains both DNA binding subdomains and the N-terminal protein-protein interaction domain stimulated DNA polymerase ␣. Therefore loss of this activity cannot explain the failure of RPA m in DNA replication.
Except ⌬222-411 tRPA, all the other RPAs stimulate polymerase ␦ on a primed M13 ssDNA template (Fig. 4B) to a similar extent. In fact, both 278 -616 tRPA and RPA m consistently stimulate polymerase ␦ better than wild type RPA. The resulting products showed no differences in length when analyzed on a 1% alkaline agarose gel (data not shown). The results are compatible with the notion that most single-stranded DNAbinding proteins can stimulate mammalian DNA polymerase ␦. Since ⌬222-411 tRPA lost the high affinity DNA binding domain, it cannot stimulate polymerase ␦ as well as others. Formation of the Earliest Okazaki Fragments-Synthesis of an RNA primer and the first Okazaki fragment is carried out by the DNA polymerase ␣-primase complex following unwinding of the origin. Most of the Okazaki fragments detected in the denaturing agarose gel in Fig. 1A are synthesized during the elongation phase of a 60-min reaction. If DNA synthesis is initiated (but the products are not significantly elongated) one expects to detect the initial Okazaki fragments in a short pulse-labeled SV40 DNA replication reaction (53).
To reduce the background of T Ag-independent labeling of DNA containing single strand breaks, reaction mixtures containing different RPA mutants were incubated at 37°C for 45 min prior to addition of T Ag (68). These were then incubated with T Ag for 15 min to allow the assembly of preinitiation complex (69), which facilitated the efficiency of pulse labeling. A 1-min pulse with radiolabeled nucleotide was then provided, and the reaction was stopped. In experiments published by others, a 90-s pulse produced nascent DNA with an average size distribution of 1000 bases (53). Thus 1 min should allow the formation of the earliest Okazaki fragments. The products labeled during a 1-min pulse was analyzed on a 10% acrylamide gel containing 8 M urea (Fig. 5). Under these conditions, products from RPA-depleted cell lysates supplied with either wild type RPA (1-616 tRPA) (lane 1) or the zinc finger mutant (RPA m ) (lane 3) could easily be detected. Thus the zinc finger mutant RPA m retained the capacity to promote the formation of at least the initial Okazaki fragments of Ͼ200 nucleotides. These initial Okazaki fragments were not detected in the 0.8% alkaline agarose gel in Fig. 1 because they were few compared with the large number of Okazaki fragments produced in the reaction with wild type RPA due to extensive replication fork movement. On the other hand, when no exogenous RPA was present (lane 6), or when the other two RPA mutants were added (lanes 4 and 5), even the earliest Okazaki fragments were not detected (Fig. 5). ⌬222-411 tRPA and 278 -616 tRPA failed in the SV40 replication reaction at the earliest possible step, namely origin unwinding. Even though these mutant proteins bind single-stranded DNA (in the latter case even with high affinity), they either do not bind to DNA in the appropriate mode or are inactivated by their large deletions in an unknown function that is essential for origin unwinding. RPA m , however, allows origin unwinding and formation of the first Okazaki fragments of Ͼ200 bases but fails in a subsequent step that is essential for strand elongation.
Wild type RPA, which promoted the formation of multiple Ͼ200-base Okazaki fragments, also produced a distinct short product of 20 nucleotides which was not seen with RPA m (Fig.  5). The 20-nucleotide nascent DNA product could be the result of a pause during elongation of Okazaki fragments that corre- sponds to the switching of polymerase from DNA polymerase ␣ to polymerase ␦. The diminution of short-stop DNA corresponding to such a pause in reactions supported by RPA m could imply that polymerase switching is inefficient in this reaction. This could account for the failure to synthesize any leading strand product and lagging strand product beyond the first few Okazaki fragments.
The Zinc Finger Mutant, RPA m , Is Competent in Nucleotide Excision Repair-The activities described so far for the zinc finger mutant RPA m are intriguing. Although not able to form detectable replication products, it is competent in a variety of related functions and replication sub-reactions. The zinc finger is dispensable for DNA binding, origin unwinding, and initiation of DNA replication but appears to be required for elongation of DNA replication products. It was of interest to determine whether the zinc finger and deletion mutants could function in DNA repair. We first tested their activity in nucleotide excision repair.
Nucleotide excision repair DNA synthesis reactions were carried out with UV-irradiated and non-irradiated templates. Equal amounts of template were present as shown in the ethidium bromide-stained gel (Fig. 6A, upper panel), where UV-irradiated versus non-irradiated templates can be discriminated by their different sizes. No incorporation of radiolabeled dNTP (repair synthesis) is detected in non-irradiated template. A template-specific repair on the irradiated template is seen with both wild type protein 1-616 tRPA (lower panel, lanes 2-4) and zinc finger mutant RPA m (lower panel, lanes 5-7).
RPA m has about 50 -60% the activity of wild type RPA (data not shown). None of the other two mutants 278 -616 tRPA (lanes 8 -10) and ⌬222-411 tRPA (lanes [11][12][13] showed specific repair products compared with the background (lane 1). These data and results from an independent experiment with further increased concentration of tRPA are quantified in Fig. 6B. Residues within the region 1-278 of Rpa1 are therefore clearly needed for the full nucleotide excision repair reaction, but in contrast to SV40 DNA replication, the zinc finger is dispensable.
Since RPA is required at the incision step of nucleotide excision repair, we specifically examined this step by analyzing the incision intermediates produced in reactions containing the various RPA mutants. A closed circular DNA containing a single 1,3-intrastrand d(GpTpG)-cisplatin cross-link was used as a template for analysis. This adduct is an excellent substrate for the nucleotide excision repair system (70). The principal 3Ј incision is at the 9th phosphodiester bond from the 3Ј-platinated guanine and the principal 5Ј incision at the 16th phosphodiester bond on the 5Ј side of the lesion. Excision products 24 -32 nucleotides in length are formed, with a 26-mer product predominating (60).
A Southern hybridization procedure was used to detect the products of incision reactions with different mutant forms of RPA (Fig. 7). No excised fragments were detected when a non-platinated con-GTG DNA was used as a template (lane 2).  3 and 7) gave excised fragments with the normal size range. Apart from a minor decrease in activity, RPA m showed no qualitative difference from wild type RPA in supporting nucleotide excision repair. The other two mutants fail at the earliest step of the reaction, even though one of them (278 -616 tRPA) binds to singlestranded DNA with high affinity. The same results were seen in independent reactions where PCNA was added to reaction mixtures (data not shown), an experiment done in case PCNA was needed to stimulate incision. It is probable that the deletions affect the ability of Rpa1 to interact with other compo- nents of the nucleotide excision repair machinery. We also checked for any dominant negative effect of the Rpa1 deletion mutants by adding up to 4 g of 278 -616 tRPA or ⌬222-411 tRPA to repair reactions containing 3 g of wild type RPA. No inhibition of the reactions was found, but addition of 0.5-2 g of RPA m stimulated reactions containing wild-type RPA, consistent with the activity of the zinc finger mutant (data not shown).
Wild Type RPA but None of the Mutant Forms of RPA Rescue Mismatch Repair in RPA-depleted Cell Lysate-In E. coli, methylation is used as a strand discrimination system to choose the template in mismatch repair (71). In higher eukaryotes, it is still unclear how cells decide which strand is to be repaired in a mismatch. However, by introducing a nick, one can artificially bias the in vitro system to correct the strand with a break (72,73). We used a circular heteroduplex template containing a site-specific strand-specific nick and a single base (G/T) mismatch to examine the role of RPA in mismatch repair (Fig. 8A). The presence of the mismatch within overlapping recognition sites for two restriction endonucleases (XhoI and NsiI) renders the DNA resistant to digestion by either enzyme. Correction of the mismatch permits the assessment of strand specificity of repair. Thus the nick-driven repair will use the circular strand as template and repair the nicked strand thereby restoring the XhoI site and eliminating the NsiI site. Repair using the opposite strand as template will, on the other hand, restore the NsiI site and eliminate the XhoI site. Thus the repair process can be monitored by linearizing the substrate DNA with AlwNI, digesting with either XhoI or NsiI, and measuring the formation of a smaller fragment of 2.1 kb that is defined by the largest distance between the AlwNI site and the NsiI/XhoI sites.
For the mismatch repair assay we first used a cell lysate depleted of RPA by selective ammonium sulfate precipitation and supplemented it with recombinant RPAs. However, this lysate did not support mismatch repair even when supplied with wild type RPA probably because key mismatch repair proteins are depleted along with RPA during the ammoniumsulfate precipitation. Hence we switched to anti-RPA antibody to specifically inhibit endogenous RPA in the mismatch repair reaction. As shown in Fig. 8B, the addition of 2 g of anti-RPA Ab (␣Rpa1, 70-9) decreased both the nick-driven repair by 60% and non-nick-driven DNA repair by approximately 65%. On the other hand, the addition of 2 g of anti-hemagglutinin Ab decreased both nick-and non-nick-driven DNA repair by only 20%. Similar results were observed with another neutralizing Ab against RPA (␣Rpa1, . No repair occurred when cell lysates were not added. Since hMSH2 and hMLH1 are known to be essential in the mismatch repair (74), to further prove that the nick-driven repair is catalyzed by the MSH2-and MLH1-dependent mismatch repair pathway, monoclonal antibodies against the MSH2 and MLH1 proteins were tested for their ability to inhibit the reaction. Antibodies against MSH2 and MLH1 specifically inhibited this repair reaction (data not shown). Hence the mismatch repair reaction observed is indeed repaired by MSH2-and MLH1-dependent mismatch repair.
We next tested if RPA and mutants could reverse the inhibition due to immunodepletion of wild type RPA. The results are shown in Fig. 9A and quantitated in Fig. 9B. Although the addition of 4 g of wild type RPA (lane 4) completely rescued the inhibition by 2 g of anti-RPA Ab, the addition of the same amount of zinc finger mutant RPA m (lane 3) or other mutants (lanes 1 and 2) did not reverse the inhibition. All of the mutant RPA complexes as well as wild type RPA were recognized by the anti-RPA antibody (data not shown). Therefore, the difference observed cannot be attributed to the antibody being selec-tively neutralized by wild type but not mutant RPA complexes. This is the first demonstration that RPA plays a role in mismatch repair in human cell extracts. The larger deletions of Rpa1 permitted binding to single-stranded DNA but still inac- FIG. 8. A, map of the circular heteroduplex template showing the location of the mismatches used for the repair synthesis, the sitespecific nick at the ScaI site created during construction, and the XhoI, AlwNI sites used for repair detection. B, inhibition of mismatch repair in a cell lysate by antibody against RPA. Indicated amounts of antibodies (2 g in a 25 l reaction) were added, and the repair was followed by restriction digestion (see "Materials and Methods"). The percentage of repaired product (containing the XhoI site) relative to total product (linear plasmid) indicates nick-driven repair, whereas the percentage of product containing the NsiI site relative to total plasmid indicates non-nick-driven repair. Anti-hemagglutinin (HA) antibody was used as a negative control. tivated the protein for mismatch repair. Furthermore, unlike nucleotide excision repair but like DNA replication, mismatch repair requires an intact zinc finger in Rpa1. DISCUSSION In this study, we investigated the function of RPA in DNA repair and replication using three recombinant RPA mutants as well as wild type RPA. Among the three mutants, the zinc finger mutant is particularly interesting. Zinc finger structures are known to be important in DNA-protein and protein-protein interactions (75,76). The zinc finger is highly conserved among different RPA homologs, implying that it is important for some of the functions of RPA.
The studies presented here demonstrate that the zinc finger point mutant is able to physically interact with T Ag, assist origin unwinding, stimulate DNA polymerase ␣ and ␦ in DNA synthesis, and support the formation of the earliest Okazaki fragments. Neither of the other two mutants retained any of the functions described above except for interacting with T Ag and polymerase ␦ stimulation. The likeliest explanation for why RPA m is inactive in SV40 DNA replication but active in the various steps up to early Okazaki fragment formation is that the mutant can support the events related to initiation but not elongation. Since RPA m is able to stimulate polymerase ␦ on a primed M13 ssDNA template, the deficit could be in the loading of polymerase ␦ at replication forks. The pulse labeling experiment allowed us to examine the smallest initial Okazaki fragments. Our hypothesis is supported by the fact that a strong-stop nascent DNA product of 20 bases that could correspond to a pause associated with polymerase switching is not seen in the RPA m supported reactions. The situation is similar to a reaction without PCNA when only 1-2% DNA synthesis is seen and most of it is due to formation of short Okazaki fragments confined to the origin of DNA replication (5).
A less likely explanation for the failure of RPA m to support DNA replication is that the slightly decreased activity of this mutant protein in several of the sub-reactions add up to the complete absence of detectable products in the replication reaction. The nearly wild type activity of RPA m in nucleotide excision repair argues for the first explanation that the protein is totally inactive in a specific elongation step required for DNA replication, since elongation steps would not likely be required for excision repair.
Based on the results of SV40 DNA replication, it was speculated that RPA m (since it executes all steps up to Okazaki fragment formation) may be a dominant negative inhibitor of wild type RPA in a competitive assay. However, we did not detect any dominant negative effect of RPA m over wild type RPA. It is conceivable that RPA m fails to enter into an initiation complex in the presence of wild type RPA accounting for the absence of a dominant negative effect. Future experiments will address this issue.
In addition to DNA replication, nucleotide excision repair and mismatch repair assays were performed to assess further the functional properties of the mutant RPA proteins studied here. Among other things, the results presented here provide the first demonstration of the biochemical requirement of RPA in the in vitro mismatch repair reaction. In addition, none of the mutant forms of RPA studied here were capable of supporting mismatch repair. This was in contrast to nucleotide excision repair, which is also RPA-dependent. In this case, RPA m was able to support nucleotide excision repair in vitro, whereas neither of the deletion mutants were able to support nucleotide excision repair. In this regard RPA m exhibited an intriguing phenotype. It is unable to replace wild type RPA in mismatch repair or in DNA replication but is fully competent in nucleotide excision repair. The zinc finger area may be important for certain interactions but not others; for example, it could be essential for recruiting mismatch repair proteins and replication factors but not nucleotide excision repair proteins. DNA polymerase ␣ has been implicated in mismatch repair in addition to DNA polymerase ␦ (73,77,78). It is likely that the long stretch of DNA formed in DNA replication and mismatch repair requires switching from polymerase ␣ to ␦ and hence is affected by the mutation in the zinc finger motif of Rpa1. Filling in of a short gap during nucleotide excision repair could be executed by a single DNA polymerase, especially polymerase ⑀ and so is still supported by RPA m . Alternatively, RPA may be required in nucleotide excision repair to displace the short excised fragment produced by dual incision rather than to support DNA synthesis, whereas in mismatch repair and DNA replication RPA may be required at the elongation stage of DNA synthesis.
Our studies and those from Wold's group (20) have consistently failed to detect replication products from RPA-depleted cell lysate supplemented with the zinc finger mutant or the other two DNA domain deletion mutants. However, another group reported that RPA with a deletion of 50 amino acids including the zinc finger motif was functional in SV40 DNA replication but inhibited both DNA polymerase ␣ and ␦ (79). The discrepancy between these two results could be because the two mutants are not identical; ours is a point mutation, whereas the other is a 50-residue deletion across the zinc finger. Perhaps the existence of the unfolded zinc finger domain in RPA m is more deleterious to protein function than the absence of the domain in the deletion derivative (20).
The two RPA deletion mutants studied here bound to single- FIG. 9. A, inhibition of mismatch repair by anti-RPA antibody is reversed by 4 g of wild type RPA (1-616 tRPA) but not by any of the mutant forms of RPA in a 50-l reaction. The nick-driven mismatch repair restores an XhoI site so that digestion with AlwNI and XhoI produces a 2.1-kb DNA fragment as opposed to the linearized parental DNA of 2.9 kb (see Fig. 8A). B, quantitation of the mismatch repair reactions. The extract alone repaired 10 -15% of the DNA, and this is represented as 100% repair. Addition of 2 g of anti-RPA antibody inhibited the repair to 10 -20% of extract alone. This inhibition was rescued by 4 g of wild type RPA but not by any of the mutant forms of RPA.
stranded DNA under physiological salt concentrations, and ⌬278 -616 tRPA bound to DNA with high affinity (8). Both retained sufficient secondary structure to allow interaction with Rpa2 and Rpa3 to form the RPA holocomplex and support interaction with T Ag. However, both were inactive in the repair and replication assays. Perhaps the deletions in these Rpa1 molecules affect their mode of interaction with singlestranded DNA or their interaction with replication or repair proteins resulting in this null phenotype. An interesting feature of the data on these two deleted forms of RPA is they suggest that single-stranded DNA binding per se is not sufficient for RPA to promote the excision step in nucleotide excision repair.
The mutation analysis presented here provides strong evidence that RPA is not merely a single-stranded DNA-binding protein or a complex that brings Rpa2 and Rpa3 to the replication machinery. The failure of Rpa1 alone to support DNA replication despite binding DNA well (7) could be explained by Rpa2 and Rpa3 executing unknown but essential functions. However, all the mutant forms of RPA characterized here contain Rpa2 and Rpa3 and thus bring these subunits to regions of single-stranded DNA. This is particularly relevant in the case of RPA m since it supports origin unwinding, stimulates DNA polymerase ␣ and ␦, as well as supports DNA replication up to the formation of the first Okazaki fragments suggesting that the molecule binds to single-stranded DNA appropriately. Consistent with this, RPA m also supports nucleotide excision repair. However, its failure in replication beyond the initiation of DNA synthesis and its failure in mismatch repair strongly suggest that RPA is also required as a molecular matchmaker, coordinating activities of proteins involved in strand elongation or DNA repair and that the zinc finger region of Rpa1 could be important in this regard.