Role of the Rad1 and Rad10 Proteins in Nucleotide Excision Repair and Recombination*

In Saccharomyces cerevisiae , the RAD1 and RAD10 genes are involved in DNA nucleotide excision repair (NER) and in a pathway of mitotic recombination that occurs between direct repeat DNA sequences. In this paper, we show that purified Rad1 and Rad10 interact with a synthetic bubble structure and incise the DNA at the 5 (cid:42) -side of the centrally unpaired region. When Rad1-Rad10 and purified XPG protein (the human homolog of yeast Rad2 protein) were co-incubated with the DNA substrate, we observed incisions at both ends of the bubble. This reaction mimics the dual incision step in nucleotide excision repair in vivo . In addition, the re- cent suggestion that Rad1 can act to resolve Holliday junctions (Habraken, Y., Sung, P., Prakash, L., and Prakash, S. (1994) Nature 371, 531–534), explaining the recombination defect observed in rad1 mutants, has been further investigated. However, using proteins purified in two different laboratories we were unable to show any interaction between Rad1 and synthetic Holliday junctions. The role that Rad1-Rad10 plays in re- combination is likely to resemble its activity in NER by acting upon partially unpaired DNA intermediates such as those formed by recombination mechanisms involv-ing single-strand DNA annealing.

In Saccharomyces cerevisiae, the RAD1 and RAD10 genes are involved in DNA nucleotide excision repair (NER) and in a pathway of mitotic recombination that occurs between direct repeat DNA sequences. In this paper, we show that purified Rad1 and Rad10 interact with a synthetic bubble structure and incise the DNA at the 5-side of the centrally unpaired region. When Rad1-Rad10 and purified XPG protein (the human homolog of yeast Rad2 protein) were co-incubated with the DNA substrate, we observed incisions at both ends of the bubble. This reaction mimics the dual incision step in nucleotide excision repair in vivo. In addition, the recent suggestion that Rad1 can act to resolve Holliday junctions (Habraken, Y., Sung, P., Prakash, L., and Prakash, S. (1994) Nature 371, 531-534), explaining the recombination defect observed in rad1 mutants, has been further investigated. However, using proteins purified in two different laboratories we were unable to show any interaction between Rad1 and synthetic Holliday junctions. The role that Rad1-Rad10 plays in recombination is likely to resemble its activity in NER by acting upon partially unpaired DNA intermediates such as those formed by recombination mechanisms involving single-strand DNA annealing.
The process of nucleotide excision repair plays a major role in the removal of DNA lesions following DNA damage. In bacteria, excision repair is initiated by three proteins, UvrA, UvrB, and UvrC, which promote a dual incision reaction by cleaving the 3rd to 5th phosphodiester bond 3Ј to the lesion and the 8th bond at the 5Ј-side of the lesion. The 12-13-nucleotide-long oligomer containing the lesion is then removed from the DNA, and the single-stranded gap is filled and sealed by DNA polymerases and DNA ligase (1,2). Dual incision is also observed in eukaryotes, and higher organisms excise a fragment of 27-29 nucleotides in length (3).
The isolation of UV-sensitive mutants has helped to define the genetic complexity of nucleotide excision repair in Saccharomyces cerevisiae, and it is now known that many genes are involved in the process. These include RAD1, RAD2, RAD3,  RAD4, RAD7, RAD10, RAD14, RAD16, RAD23, TFB1, SSL1, and SSL2 (1,4). Remarkably, the sequences of excision repair genes in yeast and mammalian cells are highly homologous, and the enzymology of the two systems is very similar. The RAD1 and RAD10 genes of S. cerevisiae encode polypeptides that interact to form a stable complex (5-7) that exhibits a structure-specific endonuclease activity. The endonuclease cleaves single-stranded DNA tails with 3Ј-ends at the junction with duplex DNA (8), indicating that Rad1-Rad10 promotes the 5Ј-incision event in NER. 1 Genetic studies show that rad1 and rad10 mutants also exhibit defects in mitotic recombination (9 -14). In particular, RAD1 and RAD10 are required for intrachromosomal recombination between direct repeats and affect the integration of linear DNA molecules and circular plasmids into genomic sequences. RAD1 is also required for doublestrand break repair between two direct repeats when the double-strand break occurs within a region of heterology (12).
Recently it was suggested that Rad1 can resolve Holliday junctions (15), thus explaining the recombination defect observed in rad1 mutants.
In this paper we have further investigated the specific endonuclease activity of Rad1-Rad10 using substrates that mimic intermediates thought to occur during nucleotide excision repair and recombination. We show that the purified Rad1 and Rad10 proteins cut a synthetic bubble structure at the 5Ј-side of the centrally unpaired region and that Rad1-Rad10 together with purified XPG protein (the human homolog of yeast Rad2 protein) catalyze a dual incision reaction. In contrast to the report by Habraken et al. (15), we did not detect any cleavage of Holliday junctions by Rad1 or Rad1-Rad10.

EXPERIMENTAL PROCEDURES
Proteins-The S. cerevisiae Rad1 and Rad10 proteins were purified to homogeneity as described (16). Two preparations of Rad1 were utilized (prepared independently in the laboratories of E. C. F. and A. E. T.). Rad1 protein was found to be unstable and became inactivated by repeated freeze/thawing, and consequently it was always used from a freshly thawed stock. When complexed with Rad10, Rad1 was considerably more stable. Rad1 and Rad10 were therefore preincubated together on ice for 5 min before addition to the reactions. The two preparations gave very similar results. Escherichia coli RuvC protein (17) and human XPG protein (18) were purified as described elsewhere.
DNA Substrates-The bubble substrate was produced by annealing two 90-mer oligonucleotides. One oligonucleotide contains a central poly(C) 30 region, whereas the other contains poly(T) 30 to produce a centrally unpaired poly(C)⅐poly(T) region of 30 nucleotides flanked by duplex DNA. The DNA sequences and method of preparation are described elsewhere (19). The poly(T) 30 -containing strand (oligo 2) was labeled with 32 P at the 5Ј-end prior to annealing. Synthetic Holliday junction X12 was prepared as described previously (20).
Cleavage Assays-Reaction conditions are stated in the figure legends. 32 P-Labeled DNA products were analyzed by denaturing 12% polyacrylamide gel electrophoresis using TBE buffer. The gels were dried onto filter paper, exposed to Kodak XAR5 film, and visualized by autoradiography.

RESULTS AND DISCUSSION
A key event in NER in eukaryotic cells is the introduction of dual incisions flanking a damaged site in DNA, leading to the * This work was supported by the Imperial Cancer Research Fund (to S. C. W. and R. D. W.), the National Institutes of Health (to E. C. F.), and the Council for Tobacco Research (to A. E. T.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ To whom correspondence should be addressed.  (3). To investigate the incision step in more detail, we prepared a synthetic bubble structure, which contains a 30-nucleotide-long unpaired region flanked by duplex regions of similar length (Fig. 1). This model substrate is likely to resemble an intermediate in NER, after repair proteins have unwound and separated the DNA strands around the damaged region (19). When purified S. cerevisiae Rad1 and Rad10 proteins were incubated with a synthetic bubble structure in which one strand was 5Ј-32 P-end labeled, we observed cleavage of the labeled strand at the 5Ј-side of the unpaired region, ϳ2 bases into the adjacent duplex DNA (Fig. 1, lanes b and c). No products were observed that corresponded to cleavage within the unpaired region or on the 3Ј-side. The polarity of cleavage is therefore the same as that observed previously with model DNA structures with single-stranded DNA tails with 3Ј-ends (8). However, the results presented here provide evidence that a free single-stranded DNA end is not required for structure-specific cleavage by Rad1-Rad10, indicating that the duplex ␦-single-strand junction is directly recognized by the endonuclease. Cleavage of the bubble structure required both proteins (Fig. 2, lane d), since neither Rad1 (lane b) nor Rad10 (lane c) alone had any effect on the bubble substrate.
In human cells, the XPG endonuclease makes the 3Ј-incision during nucleotide excision repair (19). When XPG was compared with Rad1-Rad10, we observed that the bubble substrate was nicked with a polarity opposite to that seen in reactions with the Rad1-Rad10 complex (Fig. 1, lanes e and f). When the bubble substrate was incubated with both XPG and the Rad1-Rad10 protein complex (Fig. 1, lane d), incision products were observed that corresponded to cleavage at the 3Ј-and 5Ј-sides of the bubble.
The S. cerevisiae homologue of the human XPG protein is Rad2 (21). Consistent with this, a truncated form of Rad2 has been shown to cleave model DNA structures with the same polarity as XPG (22). The dual incision reaction of NER in yeast is therefore likely to involve Rad2 and the Rad1-Rad10 complex, with the corresponding functions performed in humans by XPG and the ERCC1-ERCC4 protein complex (where ERCC1 is the homologue of Rad10 (23) and ERCC4 is the homologue of Rad1 2 ). Although the exact mechanics of damage recognition, DNA unwinding, and incision remain to be elucidated, the results presented here strengthen the view that a partially unwound or open complex around a DNA lesion is an intermediate in the NER process.
Genetic evidence indicates that, in addition to their role in NER, the Rad1 and Rad10 proteins are involved in a pathway of mitotic recombination (1,4), possibly by playing a role in the resolution of recombination intermediates (such as Holliday junctions). Recent studies reported that the Rad1 protein (1-5 g/ml) could resolve synthetic Holliday junctions by endonucleolytic cleavage (15). Since the structure of a Holliday intermediate is quite different from intermediates that occur during NER, we compared the activities of Rad1 and Rad1-Rad10 with RuvC protein, the E. coli Holliday junction resolvase (24,25). Fig. 2 (lane d) shows that Rad1-Rad10 cut the bubble structure efficiently whereas RuvC protein was inactive on this substrate (lane e). We also compared the activities of Rad1-Rad10 and RuvC on a synthetic Holliday junction X12 (20). Resolution by RuvC is known to occur by the introduction of symmetrically related nicks at the site of the crossover (20). When the junction was 5Ј-32 P-labeled in strand 2, we observed the nick in this strand by denaturing polyacrylamide gel electrophoresis (Fig. 3, lane h). However, using preparations of Rad1 produced in two different laboratories, we were unable to detect any evidence of nicking when the junction was incubated with 1.2-12.5 g/ml purified Rad1 protein (Fig. 3, lanes b-e,  and data not shown). Incubation of the junction with the Rad1-Rad10 complex also failed to yield cleavage products (Fig. 3,  lane g). Experiments carried out with a different junction (X0 (20)) also provided negative results (data not shown). Furthermore, we could find no evidence that Rad1 or Rad10 could bind to the synthetic Holliday junctions, either individually or as a  (16) and for XPG were 20 mM Hepes, pH 6.8, 10% glycerol, 2 mM MgCl 2 , 1 mM dithiothreitol, and 50 g/ml BSA (lane f) (18). All other reactions (lanes c, d, and e) were carried out in a compromise buffer (50 mM Tris-HCl, pH 7.5, 20 mM NaCl, 10% glycerol, 5 mM MgCl 2 , 1 mM dithiothreitol, and 50 g/ml BSA). Amounts of proteins were: Rad1 (100 ng), Rad10 (20 ng), and XPG (100 ng). Reaction mixtures (20 l) were incubated for 90 min at 37°C, and 32 P-labeled DNA products were analyzed by 12% denaturing polyacrylamide gel electrophoresis and visualized by autoradiography. Lane g, oligonucleotide markers of defined length.

Role of Rad1 in Repair and Recombination 24639
complex, as measured by bandshift assays (data not shown). Taken together, these results show that under conditions in which the Rad1-Rad10 complex is highly active and cleaves 85% of the bubble structure (Fig. 3, lane i), it fails to nick the synthetic Holliday junction in a manner expected of a junctionspecific resolvase. Though we cannot account for the ability of others to demonstrate Holliday junction resolution by Rad1 (15), it is known that synthetic Holliday junctions tend to "breathe" during incubation and may be prone to attack by contaminating single-strand endonucleases (26).
If Holliday junctions do not serve as targets for Rad1-Rad10, what role do these proteins play in recombination? Studies of budding yeast indicate that there are two pathways of doublestrand break-induced recombination (27). One of these pathways (single-strand annealing or SSA) is utilized during recombination between repeat sequences (Fig. 4B), and deletions between repeats are greatly reduced in the absence of either the RAD1 or RAD10 gene products (9, 13, 14, 28 -30). However, the SSA pathway for recombination is unlikely to proceed via a Holliday junction intermediate. An important clue to the role of Rad1 in recombination was provided by Fishman-Lobell et al. (12), who observed that rad1 mutants were defective in recombination between two direct repeat sequences when one of the repeats contained a heterologous region in which recombination was initiated by a double-strand break. It was suggested that RAD1 was required for an endonucleolytic activity that acts in the removal of non-homologous 3Ј-single-stranded ends at an early stage of recombination (12). An identical defect in double-strand break-induced recombination is also present in rad10 mutants but not in mutants of other NER genes (29). An absence of Rad1 or Rad10 would thus block SSA between direct repeats.
Duplex/3Ј-single-strand DNA junctions that model those thought to be present during the SSA pathway (12) are known to be acted upon by the Rad1-Rad10 nuclease (Ref. 8 and this work). The in vitro properties of Rad1-Rad10, indicative of a structure-specific nuclease, can therefore simultaneously account for a role in the cleavage of some recombination intermediates (Fig. 4B), although not Holliday junctions, as well as the specific cleavage of DNA nucleotide excision repair intermediates (Fig. 4A). The unwound region (modeled in the experiments described here by a synthetic bubble structure) is recognized by the Rad1-Rad10 complex and Rad2, which introduce structure-specific incisions at the 5Ј-and 3Ј-sides, respectively. In mammalian cells, similar cuts would be made by ERCC1-ERCC4 complex and XPG, respectively. Following incision and removal of the damage-containing oligonucleotide, the gap would be filled by DNA resynthesis. B, model for the recombinational repair of double-strand breaks by SSA. Following the initiation of recombination by the introduction of a double-stranded break, the broken DNA ends are resected by the action of a 5Ј-3Ј-exonuclease. Complementary overlapping 3Ј-ends then anneal, and Rad1-Rad10 trims off the 3Ј-ended unpaired single strands. The DNA is then repaired by resynthesis and ligation, resulting in a deletion product in which sequences between two direct repeats have been lost.