Structure-specific nuclease activity in yeast nucleotide excision repair protein Rad2.

Saccharomyces cerevisiae Rad2 protein functions in the incision step of the nucleotide excision repair of DNA damaged by ultraviolet light. Rad2 was previously shown to act endonucleolytically on circular single-stranded M13 DNA and also to have a 5'-->3' exonuclease activity (Habraken, Y., Sung, P., Prakash, L., and Prakash, S. (1993) Nature 366, 365-368; Habraken, Y., Sung, P., Prakash, L., and Prakash, S. (1994) J. Biol. Chem. 269, 31342-31345). Using two different branched DNA structures, pseudo Y and flap, we have determined that Rad2 specifically cleaves the 5'-overhanging single strand in these DNAs. Rad2 nuclease is more active on the flap structure than on the pseudo Y structure. Rad2 also acts on a bubble structure that contains an unpaired region of 14 nucleotides, but with a lower efficiency than on the pseudo Y or flap structure. The incision points occur at and around the single strand-duplex junction in the three classes of DNA structures.

A mammalian 45-kDa 5Ј 3 3Ј exonuclease that is related to E. coli pol I 5Ј 3 3Ј nuclease and that shares homology with yeast Rad2 is required for lagging strand DNA synthesis in reconstituted DNA replication systems (6 -10). Following the RNase H1-catalyzed cleavage of primer RNA one nucleotide 5Ј of the RNA-DNA junction, the 5Ј 3 3Ј exonuclease removes the remaining monoribonucleotide of the RNA primer (9). Like the pol I nuclease, the mammalian enzyme also has a similar structure-specific activity (11). The RTH1 gene encodes the S. cerevisiae counterpart of this mammalian 45-kDa exonuclease (12). Genetic studies with the rth1⌬ mutant strain have indicated a role of RTH1 in DNA replication as well as in DNA mismatch repair (12,13).
The protein encoded by RTH1 and its mammalian counterpart contains ϳ380 amino acids. Rad2, by contrast, is a much larger protein, containing 1031 residues. The homology between yeast Rad2, RTH1, and their mammalian counterparts is restricted to three regions (for references, see Ref. 12). Moreover, whereas RTH1 and its mammalian counterparts have a role in DNA replication and in mismatch repair, Rad2 is required for NER, but has no apparent involvement in DNA replication and mismatch repair (2,3,12). 2 Thus, Rad2 and RTH1 proteins have diverged functionally. Here, we examine the action of Rad2 nuclease on various DNA structures.
Polyclonal Antibodies Specific for Rad2-A ␤-galactosidase-Rad2 fusion protein was expressed in E. coli carrying the plasmid pKM8 (14). The insoluble hybrid polypeptide was purified from inclusion bodies by preparative SDS-polyacrylamide gel electrophoresis and used as antigen for polyclonal antibody production in rabbits. Antibodies were affinity-purified from the rabbit sera by passage through a Sepharose column containing the covalently linked antigen. After dialysis against phosphate-buffered saline (10 mM NaH 2 PO 4 , pH 7.2, 150 mM NaCl), the antibody solution was concentrated to 2 mg/ml and stored in small portions at Ϫ70°C. Affinity-purified polyclonal antibodies specific for Rad1 and Rad10 proteins were obtained as described previously (15).
DNA Substrates-Oligonucleotides A, C, E, and G were radiolabeled at their 5Ј-terminus with the bacteriophage T4 polynucleotide kinase and [␥-32 P]ATP (Amersham Corp.; 6000 Ci/mmol). DNA substrates were obtained by annealing one of the radiolabeled oligonucleotides to nonlabeled oligonucleotides. Radiolabeled oligonucleotide C was hybridized to oligonucleotide A to generate the pseudo Y-1 substrate shown in Fig. 1, and radiolabeled oligonucleotide C was hybridized to oligonucleotides A and B to generate Flap-1, which contains a 5Јoverhanging single strand (see Fig. 1). Radiolabeled oligonucleotide E was hybridized to oligonucleotide A and radiolabeled oligonucleotide E was hybridized to oligonucleotides A and D to generate the pseudo Y-2 and 3Ј-overhanging Flap-2 DNA substrates, respectively (see Fig. 1). To obtain the bubble DNA substrate (see Fig. 1), radiolabeled oligonucleotide G and oligonucleotide F were hybridized to each other. Annealing reactions were carried out by mixing 180 pmol of radiolabeled oligonucleotides with 200 pmol of nonradiolabeled oligonucleotides in 50 l of buffer (50 mM Tris-HCl, pH 8.2, 10 mM MgCl 2 , 0.2 mM EDTA, 5 mM dithiothreitol, and 0.1 mM spermidine) and incubating at 92°C for 2 min, at 65°C for 10 min, at 37°C for 20 min, and finally at 25°C for 20 min. Annealing mixtures were run at 4°C on 8% polyacrylamide gels in TBE buffer (90 mM Tris borate, pH 8.5, 2 mM EDTA); the region of the gel containing the radiolabeled DNA substrate was excised; and the DNA was eluted by diffusion from the crushed gel slice at 4°C overnight into TE buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 50 mM KCl). The purified DNA substrates were stored at 4°C.
Nuclease Assay-Rad2 protein (2-40 ng or 17-340 fmol) was incubated with 75 fmol of 5Ј-32 P-labeled DNA (7,000 -20,000 cpm) in reaction buffer (50 mM Tris-HCl, pH 8.0, 5 mM MgCl 2 , 1 mM dithiothreitol, and 100 g/ml bovine serum albumin) for 10 min at 30°C in a final volume of 11 l. The reaction was terminated by adding 11 l of gel loading buffer (90% deionized formamide in TBE buffer); the mixture was heated for 2 min at 92°C, and a fraction of it (2,500 cpm) was loaded on an 11% polyacrylamide gel containing 7 M urea and 16% formamide in TBE buffer. Electrophoresis was carried out for 2 h at 30 V/cm on 0.5-mm-thick polyacrylamide gels that had been prerun at the same constant voltage for 30 min. Gels were soaked in a mixture of 20% methanol and 5% acetic acid for 15 min before being dried. The gel was either subjected to autoradiography to visualize the radiolabeled substrates and products or analyzed in the PhosphorImager 425 (Molecular Dynamics, Inc.) to quantify the various radiolabeled DNA species. The size markers used were purchased from Pharmacia Biotech Inc. and were 5Ј-end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase.

Specific Cleavage of Pseudo Y DNA Structure by Rad2
Protein-To determine whether the Rad2 nuclease activity would mediate the cleavage of a DNA structure that contains singlestranded tails adjacent to a duplex region, we hybridized 32 Plabeled oligonucleotide C to partially homologous and nonradiolabeled oligonucleotide A (see "Materials and Methods") to give the pseudo Y-1 substrate shown in Fig. 1. Reaction mixtures that contained Rad2 protein and pseudo Y-1 were incubated at 30°C and then heated for 2 min at 95°C in the presence of formamide to inactivate Rad2 protein and to separate the hybridized DNA strands. The mixture was then run, along with radiolabeled DNA size markers, on an 11% polyacrylamide gel containing 7 M urea and 16% formamide to prevent rehybridization of the DNA strands during electrophoresis. The gel was subjected to autoradiography to reveal the radiolabeled DNA species and was also analyzed in the PhosphorImager to quantify these species. The presence of DNA denaturants in the polyacrylamide gel enabled us to determine precisely the size of the cleavage products. As shown in Fig. 2 (lanes 5-8), Rad2 protein cleaved the 32 P-labeled 5Ј-overhanging single strand in pseudo Y-1 to yield two major products of 19 nucleotides (Ϸ43% of total) and 21 nucleotides (Ϸ32% of total) and two minor products of 18 nucleotides (12% of total) and 20 nucleotides (Ϸ13% of total), as determined by PhosphorImager analysis of products in lane 8. The amount of these nucleolytic cleavage products was proportional to the quantity of Rad2 added to the reaction mixture (Fig. 2, lanes  6 -8). Based on the sizes of these nucleolytic products, it can be deduced that the sites of cleavage in the radiolabeled strand in pseudo Y-1 by Rad2 protein range from 1 base into the duplex region to 2 bases away from the duplex region (see Fig. 6).
The cleavage reaction mediated by Rad2 protein is specific for the 5Ј-overhanging single-stranded tail in the pseudo Y structure because (i) no cleavage of oligonucleotide A used in the construction of pseudo Y-1 occurred at the same (Fig. 2, lane 2) and higher (data not shown) concentrations of Rad2 protein, and (ii) the 3Ј-overhanging single strand in a similar DNA structure (pseudo Y-2) that was obtained by hybridizing radiolabeled oligonucleotide E to nonlabeled oligonucleotide A (see Fig. 1 and "Materials and Methods") was not cleaved by Rad2 (Fig. 2, lane 4).
Rad2 Nuclease Cleaves Flap DNA Structure-The murine FEN-1 nuclease, a structural homolog of Rad2 protein, does not cleave pseudo Y DNA structures, but acts efficiently on "flap" DNA structures that contain a 5Ј-overhanging single strand (11), cleaving the 5Ј-overhanging single-stranded tail in the flap structures at and around the single strand-duplex junction. However, FEN-1 does not act on flap structures that contain a 3Ј-overhanging single-stranded tail (11). The results presented above indicate that, unlike FEN-1, Rad2 nuclease cleaves pseudo Y DNA efficiently, showing specificity for the 5Ј-overhanging single strand in pseudo Y (Fig. 2). To investigate whether Rad2 also cleaves flap DNA, we hybridized 5Ј-32 P-labeled oligonucleotide C to nonradiolabeled oligonucleotides A and B to form the Flap-1 substrate, which contains a 20-nucleotide 5Ј-overhanging single strand (Fig. 1), and we also hybridized 5Ј-radiolabeled oligonucleotide E to nonlabeled oligonucleotides A and D to form the Flap-2 structure, which contains a 3Ј-overhanging single strand (Fig. 1). The two flap substrates were then incubated with Rad2 protein under reaction conditions employed for the cleavage of pseudo Y-1. As shown in Fig. 3A, Rad2 protein cleaved Flap-1 highly efficiently, yielding a major product of 21 nucleotides (Ϸ74%), a lesser product of 20 nucleotides (Ϸ21%), and a trace of a product of 19 nucleotides (Ϸ5%). From the sizes of the cleavage products, it could be deduced that the major site of cleavage is located at 1 base inside the duplex region, and the minor sites are at the single strand-duplex junction and 1 base into the single-stranded region (see Fig. 6). In contrast, no cleavage of the 3Ј-overhanging single strand in Flap-2 was detected at the highest amount of Rad2 protein under the same reaction conditions (Fig. 3A, lane 6). Thus, the Rad2 flap cleaving activity closely resembles that of the FEN-1 protein in showing specificity for a 5Ј-overhanging single strand only.
The Flap-1 and pseudo Y-1 cleaving activities are intrinsic to Rad2 protein because (i) the Rad2 protein used in this study is essentially homogeneous, and (ii) the cleavage of Flap-1 (Fig.  3B) and pseudo Y-1 (data not shown) was strongly inhibited by affinity-purified antibodies raised against Rad2 protein (compare lanes 3 and 2) expressed in and purified from E. coli (14), but it was unaffected by antibodies specific for Rad1 and Rad10 proteins (lanes 4 and 5).
Rad2 Cleaves Flap Structures More Efficiently than Pseudo Y-The results presented in Figs. 2 and 3 indicate that Rad2 nuclease acts on the 5Ј-overhanging single-stranded tail in both pseudo Y and flap structures. Since Flap-1 differs from pseudo Y-1 only in possessing oligonucleotide B, which is the exact complement of the 3Ј-overhanging single strand in the latter, a direct and meaningful comparison of the relative cleavage efficiencies of the two classes of DNA structure was possible. Since 32 P-labeled oligonucleotide C was common to pseudo Y-1 and Flap-1, the two DNA structures used in the comparison had the same molar specific radioactivity. As shown in Fig. 4, the Flap-1 substrate was cleaved by Rad2 protein at a significantly higher efficiency than was pseudo Y-1. For instance, whereas Ϸ90% of Flap-1 was cleaved by 10 ng of Rad2 protein, Ϸ30% of pseudo Y-1 was cleaved by the same quantity of Rad2 (Fig. 4). The conclusion regarding the relative activities of Rad2 protein on the pseudo Y and flap structures was validated in at least three other independent experiments (data not shown).
The DNA structure-specific nuclease activity of Rad2, as assayed using Flap-1 DNA as substrate, requires Mg 2ϩ , which cannot be replaced by Ca 2ϩ , Co 2ϩ, Cu 2ϩ , or Zn 2ϩ , although Mn 2ϩ is partially effective ( Table I). The flap cleavage activity is not affected by KCl concentrations up to 50 mM, but higher amounts of the salt result in significant inhibition of the activity (data not shown), and the activity is abolished by 0.1% SDS ( Table I).
Cleavage of 5Ј-Overhanging Single Strand in "Bubble" Structure-To investigate whether Rad2 nuclease would act on a single-stranded region located in a bubble structure, we hybridized 32 P-labeled oligonucleotide G to nonlabeled oligonucleotide F to yield a bubble substrate that contains unpaired singlestranded regions 14 bases in length (Fig. 1). The bubble DNA was incubated with increasing concentrations of Rad2 protein under standard conditions, along with the same molar amount of pseudo Y-1 and Flap-1 for comparison. As shown in Fig. 5, there was a Rad2 protein concentration-dependent cleavage of the bubble DNA. The product of Rad2 nuclease action was 39 nucleotides in length, indicating that the incision site is located at 1 base into the duplex region that is proximal to the 3Ј-end of the labeled strand in the bubble DNA (Fig. 6). In this (Fig. 5) and three other independent experiments (data not shown), Rad2 protein acted with a lower efficiency on the bubble structure than on either pseudo Y-1 or Flap-1. Whether this lower efficiency is related to the DNA structure and/or is due to a DNA sequence effect is not clear at present. However, the results obtained with the bubble DNA indicate that Rad2 nuclease acts on the single strand-duplex DNA junction in the absence of a free end in the single strand. DISCUSSION Our work indicates that Rad2 cleaves flap and pseudo Y structures and that it is more active in cleaving flap structures than pseudo Y. In this regard, Rad2 resembles the mammalian FEN-1 and S. cerevisiae RTH1 nucleases, which are also more efficient at cleaving flap structures than pseudo Y (11). A variety of experiments have indicated that the E. coli pol I 5Ј 3 3Ј exonuclease gains access to the cleavage site by moving from the free 5Ј-end of single-stranded DNA to the site of cleavage at the junction with duplex DNA (5). Biochemical studies of FEN-1 and calf thymus 5Ј 3 3Ј exonuclease have indicated a similar requirement for a free 5Ј-end for strand cleavage to occur. 3 In agreement with these observations, we find no cleavage of bubble structure by the S. cerevisiae RTH1 protein that we have purified to near homogeneity (data not shown). Unlike RTH1 and FEN-1 nucleases, Rad2 cleaves the bubble structure, albeit with a lower efficiency than the flap or pseudo Y structure. The differential ability of Rad2 and RTH1 proteins to cleave bubble DNA may reflect the affinity of these proteins for binding bubble DNA. Rad2 may possess a domain that confers the ability to bind bubble DNA, and the inability of RTH1 to cleave bubble DNA may arise from the absence of this domain.
The ability of FEN-1/RTH1 to cleave 5Ј-end single-stranded DNA at its junction with duplex DNA has led to the suggestion that Rad2 cleaves the damaged DNA strand on the 3Ј-side of the damage during NER (11). While the manner of cleavage of model DNA substrates by Rad2 reported in our present work and by Rad1-Rad10 reported by Bardwell et al. (16) is congruent with the proposal that these proteins incise the damaged DNA strand on the 3Ј-and 5Ј-sides of the damage, respectively, direct evidence demonstrating this cleavage pattern in NER is as yet unavailable. Since neither the Rad2 nor Rad1-Rad10 nuclease has any affinity for damaged DNA (14,17,18), they must be targeted to the damage site via interaction with the damage recognition factors. The interaction of human XPA with the ERCC1 protein (19,20) would suggest that the Rad1-Rad10 nuclease is targeted to the damage site via interaction with the damage recognition protein Rad14. Interaction with the other components of the NER machinery may target Rad2 to the damage site. It remains to be determined whether the site of placement of the Rad1-Rad10 and Rad2 nucleases on the damaged DNA strand is coincident with the cleavage pattern of these enzymes on model DNA substrates. The recent reconsti-3 R. Bambara, personal communication.

TABLE I Effect of divalent metal ions on Rad2-mediated flap cleavage
Flap-1 DNA (75 fmol) was incubated for 10 min at 30°C with 7.5 ng of Rad2 protein and 5 mM MgCl 2 in the complete reaction or with a 5 mM concentration of the chloride salt of one of the indicated divalent metals in the absence of MgCl 2 . After electrophoresis, the cleavage products were quantified in the PhosphorImager; 100% activity corresponds to conversion of 70% of Flap-1 to products. The cleavage pattern in the presence of Mn 2ϩ was the same as that of Mg 2ϩ (data not shown).    6. Summary of cleavage sites in DNA substrates. The numbers Ϫ1, 0, ϩ1, and ϩ2 indicate cleavage sites 1 base into the duplex region, at the single strand-duplex junction, 1 base into the singlestranded region, and 2 bases into the single-stranded region, respectively. tution of the incision step of NER with purified components in yeast (3) should make it feasible to ascertain the manner of assembly and the site of cleavage by these nucleases.