Mechanism of tracking and cleavage of adduct-damaged DNA substrates by the mammalian 5'- to 3'-exonuclease/endonuclease RAD2 homologue 1 or flap endonuclease 1.

The mammalian 5′- to 3′-exonuclease/endonuclease, called RAD2 homologue 1 or flap endonuclease 1, has a unique cleavage activity, dependent on specific substrate structure. On a primer-template, in which the primer has an unannealed 5′-tail, endonucleolytic cleavage near the annealing point releases the tail intact. Entering at the 5′-end, the nuclease tracks along the entire tail to the point of cleavage. Genetic analyses suggest that this nuclease removes DNA adducts in vivo (Sommers, C. H., Miller, E. J., Dujon, B., Prakash, S., and Prakash, L. (1995) J. Biol. Chem. 270, 4193-4196). Micrococcal nuclease footprinting shows that after tracking the nuclease protects a region of the tail 25 nucleotides long, adjacent to the cleavage site. Substrates with adducts at specific locations were used to assess the mechanism of RAD2 homologue 1 nuclease tracking and its ability to cleave modified DNA. Either a conventional cis-diamminedichloroplatinum (II) (CDDP) or a bulky CDDP derivative was placed within or beyond the region protected by the nuclease. The nuclease cleaved the tail of both substrates. In contrast, a CDDP adduct just adjacent to the expected cleavage point was inhibitory. A CDDP adduct at the very 5′-end of the tail was also cleaved. The nuclease could remove tails containing adducts on the sugar-phosphate backbone. Apparently, the nuclease is designed to slide over various types of damage on single stranded DNA and then cut past the damaged site.

Many of the reactions that occur at the mammalian DNA replication fork have been reconstituted in vitro using purified enzymes, as reviewed by Bambara and Huang (1). We have studied the processing of Okazaki fragments during lagging strand DNA replication using reconstitution reactions with purified calf enzymes. We found that the initiator RNA primers were removed by the combined action of two nucleases, RNase H1 and a 5Ј-to 3Ј-exonuclease (2). RNase H1 cleaves the initiator RNA 1 residue upstream of the RNA-DNA junction, leaving a single 5Ј-terminal ribonucleotide. This remaining RNA residue is removed by the 5Ј-to 3Ј-exonuclease. Following RNA removal, the exonuclease works coordinately with any polymerase in a nick translation reaction (3) followed by ligation (4). We also found that this nuclease removes unannealed 5Ј-tails of primers on templates through endonucleolytic cleavage (5).
The mammalian 5Ј-to 3Ј-exonuclease/endonuclease has been purified and examined by several groups. The calf enzyme was first described as a double strand specific exonuclease stimulated by synthesis from an upstream primer (3). The homologous murine enzyme was isolated and named circle closing activity exonuclease (cca exonuclease) or flap endonuclease 1 (FEN1) 1 (6,7). Several groups have purified the human homologue from HeLa cells (8 -11). This enzyme was named maturation factor I (9) or DNase IV (8,10). The yeast homologue has been purified from Saccharomyces cerevisiae and named RAD2 homologue 1 (RTH1) (12,13). Analysis indicates that RTH1 shares homology with the RAD2 gene product (14), placing RTH1 with an important family of repair related nucleases. The RAD2 protein has similar cleavage activities to RTH1 in that the RAD2 nuclease removes the unannealed 5Ј-tails of primers on templates through endonucleolytic cleavage (15), as well as degrades fully annealed downstream primers through a 5Ј to 3Ј exonucleolytic activity (16). However, unlike RTH1, the RAD2 protein is capable of cleaving bubble structures. Throughout this article, we will refer to the calf and human nucleases as cRTH1/FEN1 and hRTH1/FEN1, respectively.
In addition, bacterial homologues are present in Escherichia coli DNA polymerase I and Thermus aquaticus (Taq) DNA polymerase (17). These enzymes contain domains that are functionally similar and share sequence homology with the mammalian enzymes.
Genetic analyses indicate important roles for RTH1/FEN1 nuclease in DNA replication and DNA repair. The null mutants of RTH1/FEN1 in S. cerevisiae are temperature-sensitive. At the nonpermissive temperature, mutants have the terminal cell morphology phenotype indicative of a DNA replication defect (12). Also, RTH1/FEN1 mutants have an increased sensitivity to methyl methane sulfonate, an alkylating agent, implicating RTH1/FEN1 in DNA repair (12).
Analysis of the unique substrate binding and cleavage mechanism of the RTH1/FEN1 nuclease suggests how it could remove adduct-damaged segments of DNA. Short primers annealed at different positions on the tail structure of the endonuclease substrate inhibit cleavage by RTH1/FEN1 nucle-ase (18). Avidin conjugated to biotinylated nucleotides is also inhibitory. These results indicated that RTH1/FEN1 is required to bind at the 5Ј-end and track along the tail until the endonucleolytic cleavage site is reached (18). However, not all structures are inhibitory. Biotinylated nucleotides located at different sites on the tail do not prevent cleavage. Evidently, the tracking process can tolerate some structural modification of the tail. This particular observation suggests that the role of RTH1/FEN1 in DNA repair is a direct one. RTH1/FEN1 nuclease could slide past DNA damage on an unannealed strand and removes a segment of the strand containing the damaged site. However, its motion may be blocked by some types of damage, limiting its repair capacity.
Since the discovery of cis-diamminedichloroplatinum (II) (CDDP) in 1968 as an antitumor agent, it has become one of the most important drugs for treatment of solid tumors (19). The mechanism of cytotoxicity of CDDP is thought to be exerted through the binding of the drug to DNA. Binding of CDDP induces various interstrand and intrastrand cross-links. The formation of these cross-links has been shown to inhibit replication and transcription. Although these lesions are transient, because they can be repaired (20), Sorrensen and colleagues (21) have suggested that the G 2 block caused by this damage induces apoptosis. This is likely the basis of the antitumor activity of CDDP. Currently, this agent is one of the most effective drugs available for the treatment of testicular, ovarian, bladder, head and neck, small cell lung, and cervical cancer. Unfortunately, patients eventually develop resistance to this drug, which is often accompanied by resistance to other chemotherapeutic agents (22). Resistant cell lines have increased DNA repair capabilities (23,24). Recently, it has been suggested that deficiencies in DNA repair may result in an increased sensitivity of some tumors to CDDP (25). CDDP has been shown to be removed from DNA through the nucleotide excision repair pathway (25)(26)(27). However, since CDDP does not inhibit the action of the RTH1/FEN1 nuclease, the use of these types of adducts are relevent to study the substrate requirements and the mechanism of damage removal for this enzyme. Also, the investigation of any proteins capable of participating in the removal of this drug is clinically relevant.
The following experiments test the ability of RTH1/FEN1 nuclease to translocate past sites modified by CDDP and other adducts.
Oligonucleotide Substrates-Oligonucleotide downstream primers were 5Ј-phosphorylated and radiolabeled using [␥-32 P]ATP and T4 polynucleotide kinase as per the manufacturer's instructions, with the exception of the DNA downstream primer used in the footprinting analysis. This downstream primer (see Fig. 3) was annealed to its corresponding template and extended with Sequenase version 2.0 and [␣-32 P]dCTP by the methods of the manufacturer. In all cases, the downstream primers share a region of homology between their 3Ј-end regions and the 5Ј-end regions of their respective templates. Once annealed, these primers create a substrate with an unannealed 5Ј-tail, as shown in the figures. The respective upstream primers were annealed to their templates to create a nicklike structure between the annealed regions of the two primers.
Platination of DNA Substrates-DNA oligonucleotides were platinated as described previously (28). Essentially, CDDP was incubated with gel-purified DNA in platination buffer (3 mM NaCl and 1 mM Na 2 HPO 4 ) for 18 h in the dark at 37°C. Platinated DNA was gelpurified away from nonplatinated DNA using 18% polyacrylamide, 7 M urea sequencing gel electrophoresis. Due to the slower migration of platinated DNA, purification of essentially 100% platinated DNA is guaranteed (29). We have previously demonstrated that the drug/oligomer ratio for oligomers platinated in this manner is approximately 1 (28). Platination using cis-dichloro-bis-(4-aminocyclohexanol)platinum (II) results in adducts that break down over a period of days. Therefore, experiments using this adduct contain a mixture of adducted and nonadducted DNA. However, the distinct gel mobility of adducted DNA substrates and products allowed them to be readily followed throughout our analyses.
cRTH1/FEN1 Nuclease-The 5Ј-to 3Ј-exonuclease was purified from calf thymus tissue through phosphocellulose chromatography as described previously (30). Fractions were assayed for exonuclease and endonuclease activity, using reaction conditions described below. The activities eluted in the same fractions and were pooled. The final enzyme fraction had a specific activity of 180,000 units/mg, with 1 unit being the amount of enzyme capable of hydrolyzing 1 pmol of [ 32 P]TMP from 5Ј-[ 32 P]dT 16 ⅐dA 2000 in 15 min at 37°C. hRTH1/FEN1 Nuclease-Recombinant hRTH1 was obtained from bacteria using the T7 RNA expression system (31) following the procedure of Shen et al. (32). Briefly, the coding sequence for hRTH1 was cloned into a pET expression vector, upstream and in frame with the coding sequence for 6 histidine residues. The resulting plasmid was transformed into E. coli strain BL21(DE3)pLysS (31). Transformants were grown at 37°C and induced with a final concentration of 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h. Cells were collected by centrifugation and lysed by sonication. The recombinant protein, which contains the entire amino acid sequence for hRTH1/FEN1 with a histidine tag at the C-terminal end, was purified by nickel nitrilotriacetate-agarose column chromatography as recommended by Qiagen. Active fractions were pooled and further purified by hydroxylapatite as described previously (5) for calf RTH1/FEN1. The final preparation was ϳ95% pure according to analysis by SDS-polyacrylamide gel electrophoresis stained with silver. Purified enzyme was dialyzed into a storage buffer (50% glycerol, 50 mM NaCl, 50 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, and 1 mM EDTA and EGTA) and then stored at Ϫ80°C.
Endonuclease Assay-Reaction mixtures contained 20 fmol of DNA and 2.25 units of exonuclease, 60 mM BisTris, pH 7.0, 5% glycerol, 100 mg/ml bovine serum albumin, 5 mM ␤-mercaptoethanol, and 5 mM Mg 2ϩ in a final reaction volume of 20 l. Reactions were incubated with RTH1/FEN1 nuclease and DNA substrates at 37°C for the reaction times indicated in the figure legends and terminated with 10 l of formamide dye (90% (v/v) formamide with bromophenol blue and xylene cyanole markers). Reaction products were heated to 100°C and separated by 7 M urea, 18% polyacrylamide gel electrophoresis as described by Sambrook et al. (33).
Footprinting Assay-The DNA substrate (see Fig. 3A) was 3Ј-endlabeled as described under "Oligonucleotide Substrates." 970 ng of purified hRTH1/FEN1 was incubated with 40 fmol of DNA substrate on ice for the amount of time indicated, using reaction conditions described for the exonuclease assay, with the following exception. Instead of Mg 2ϩ , which allows hRTH1/FEN1 to cleave, the reactions contained 12.5 mM CaCl 2 , which is necessary for micrococcal nuclease (MNase) cleavage activity. In this situation, hRTH1/FEN1 nuclease binds the DNA but does not cleave the substrate. MNase is able to degrade the single stranded area of the DNA not protected by hRTH1/FEN1 nuclease. After preincubating RTH1/FEN1 nuclease and DNA, the reactions were subjected to digestion by 0.075 unit of MNase for 40 s at 37°C. MNase was inactivated by the addition of 8 l of 0.25 M EDTA and EGTA, and the reaction was further processed by the addition of 17 l of formamide dye, described above.

RESULTS
Using oligonucleotide substrates, we have characterized further the mechanism of endonucleolytic cleavage by RTH1/ FEN1 nuclease. Also, we examined the types of reactions that the nuclease can perform. The substrate for endonuclease activity is a primer-template in which the downstream primer has an unannealed 5Ј-tail. Previous results show that RTH1/ FEN1 nuclease binds the downstream primer at the 5Ј-end of the tail and then tracks or slides to the point at which the primer is bound to the template. The nuclease makes an endonucleolytic cut near this junction point. Using footprint analy-sis, we have identified the region of the 5Ј-tail protected by the nuclease just prior to cleavage. In addition, we placed covalent adducts along the 5Ј-tail to determine the tolerance of the sliding process to changes in structure.
Damaged Endonuclease Substrates-Since the nuclease is obligated to fully traverse the unannealed 5Ј-tail to reach the cleavage site, structures bound to the tail may interfere with this tracking process. Although biotinylated bases conjugated to strepavidin prevent tracking by RTH1/FEN1, biotinylated bases alone permit cleavage (18). This observation suggests the mechanism whereby the nuclease could participate in repair of adduct-damaged DNA. It could slide past damage on an unan-FIG. 1. Cleavage of tails with CDDP and bulky CDDP derivative adducts. A, lanes 1-5, reactions of cRTH1 with a nonplatinated DNA control substrate. Lanes 6 -10, action of cRTH1 nuclease on a 12-nt tail that contains cis-diamminedichloroplatinum (II) adduct 6 nt from the 5Ј-end. Lanes 1 and 6, no-enzyme control lanes; lanes 2-5 and 7-10 contain enzyme. Reactions were performed for the length of time indicated under conditions described under "Experimental Procedures." B, structure of the larger cis-platinum derivative. C, reactions are the same as in A, except that lanes 6 -10 contain DNA with the cis-platinum derivative shown in B. Essentially identical results were obtained with hRTH1 nuclease. The upstream primer sequence is 5Ј-AAAAAAAACCCATTCACCACCCTGG-3Ј. The downstream primer sequence is 5Ј-ACTCCTGGCATTC-CCCTTTCCACTTTCCTCACCCCA-3Ј. The unannealed tail residues are underlined, and the platinated residues are in bold. The template sequence is 5Ј-CTGGGGTGAGGAAAGTGGAAAGGGGCCAGGGTGGTGAATGGGTTTTTTTT-3Ј. nealed 5Ј-tail and then cleave the tail, removing the damaged nucleotides. The types of damage RTH1/FEN1 nuclease may remove endonucleolytically are critical to the potential role of this enzyme in DNA repair. Previous work has not defined the range of adduct structures and sizes that allow passage of the nuclease. For our current analyses, a variety of damaged DNAtailed substrates were exposed to RTH1/FEN1 nuclease. It removes CDDP (Fig. 1A), as well as a larger CDDP derivative (Fig. 1, B and C), placed near the middle of a tail, as can be seen by the appearance of cleavage products over time. Fig. 1C shows the appearance of two sets of cleavage products. This is because the DNA substrate in lanes 7-12 actually contains a mixture of platinated and nonplatinated DNA. The larger platinum derivative does not bind as stably to the primer as does CDDP, which does not dissociate appreciably during the purification procedure. However, as can be seen in these lanes, the platinated and nonplatinated DNA are both cleaved by RTH1/ FEN1 nuclease. Also note that the DNA substrate containing the larger derivative tends to be more retarded in the gel matrix than CDDP-DNA. CDDP adducts to the N-7 of two adjacent guanine nucleotides, producing an alien structure involving two bases. Our results show that this structure does not interfere with the tracking mechanism of RTH1/FEN1 nuclease (Fig. 1A). In particular, insensitivity to the bulky CDDP adduct emphasized that the nuclease-tracking process does not involve recognition of the base structure. This could be the case if all protein-DNA contacts used for tracking occurred on the sugar-phosphate backbone. To test this postulate, we used a substrate with a tert-butyl-silyl group ( Fig. 2A) on the 2Ј-oxygen of a single ribose residue incorporated 6 nucleotides in from the 5Ј-end of a 12-nucleotide tail. Results show that the backbone adduct has no significant effect on the efficiency of sliding (Fig. 2B), as can be seen by the appearance of cleavage products over time.
Micrococcal Nuclease Footprinting-Gel shift assays (34) have detected binding of the hRTH1/FEN1 nuclease to a tailed substrate. Using the substrate in Fig. 3, we detected stable RTH1/FEN1-substrate complexes in the absence of Mg 2ϩ (data not shown). Endonucleolytic footprint analysis offered the opportunity to determine the size of the region of the unannealed 5Ј-tail protected by the nuclease. Bacterial expression of the hRTH1/FEN1 nuclease provided sufficient enzyme to carry out such an analysis. The tail length of the substrate in Fig. 3A was designed to be larger than the expected size of the region protected by the binding of the nuclease. Footprint analysis was performed in the absence of magnesium so that hRTH1/ FEN1 nuclease-directed endonucleolytic cleavage could not occur. The hRTH1/FEN1 nuclease was expected to track along the tail and bind at the position of cleavage. MNase was chosen as the footprinting nuclease because it requires calcium but not magnesium as a co-factor and can degrade single stranded DNA. As expected, a distinct footprint was observed along the tail region. The area of the substrate (depicted in Fig. 3A) protected from MNase cleavage by the presence of hRTH1/ FEN1 was approximately 25 nucleotides in length on the unannealed tail, adjacent to the point of annealing (Fig. 3). This incubation time course shows that RTH1/FEN1 nuclease tracked to the cleavage junction of the substrate quickly, by 1 min, as seen in Fig. 3A, lane 4. It stayed bound to the same location for up to 60 min (Fig. 3A, lane 13).
In view of the size of the footprint, we considered that RTH1/ FEN1 nuclease might tolerate damage near the cleavage site, because the tracking region of the protein may not have to pass that damage. For example, protein-DNA contacts essential for tracking may occur between 10 and 25 nucleotides from the cleavage site on the unannealed tail. These would not be affected by damage at positions 5 and 6. To determine whether this 5Ј-to 3Ј exonuclease/endonuclease is capable of fully traversing a damaged site, we created damaged substrates with the adducts located beyond the area protected by the nuclease in the footprinting assays. Either CDDP (Fig. 4A) or the tertbutyl-silyl group (Fig. 4B) was placed on DNA tails outside the area protected by RTH1/FEN1 in the footprinting experiments. Neither of these structures inhibited endonucleolytic cleavage, as can be seen by the appearance of cleavage products corresponding to the length of the unannealed 5Ј-tail over time. The tert-butyl-silyl group attached to the sugar-phosphate backbone in Fig. 4B causes the DNA to smear as it passes through the gel. This can be clearly seen in lanes 7-12, both above the full-length primer and above the cleavage products.
The 5Ј-End Requirement for Damaged Substrate Use by RTH1/FEN1 Nuclease-One explanation for the ability of RTH1/FEN1 nuclease to track over the adducts is that the nuclease no longer requires a free 5Ј-end in the presence of damage. To further analyze the requirement for a free 5Ј-end for entry of the RTH1/FEN1 nuclease, a "bubble" substrate was created with CDDP. The binding of a platinum adduct to a primer fully annealed to a template creates a distortion in the helix. The DNA helix is bent by 34°toward the major groove to produce local melting of 13° (35,36). This distortion disrupts base pairing, creating a bubble several nucleotides wide (37)(38)(39). The distortion in the double helix is more prevalent on the 5Ј-side of the adduct (29). Except for the melting produced by the damage, the substrate used was a fully annealed primertemplate. RTH1/FEN1 nuclease was not able to cleave any-where on the bubble substrate. This is consistent with the previous observation that DNase IV could not cleave a bubble structure created by mismatched nucleotides (8). This structure has all of the features of the RTH1/FEN1 nuclease substrate, except the free 5Ј-end. Positive control experiments were performed, demonstrating that the enzyme used was active for cleavage of a substrate with an unannealed 5Ј-tail substrate (data not shown).
CDDP Adducts at the 5Ј-End of the Tail Are Not Inhibitory-To further investigate the cleavage requirement of RTH1/FEN1 nuclease for a free 5Ј-end, we designed a substrate that contained a CDDP at the very 5Ј-end of a short tail. We considered the possibility that an adduct at this location may interfere with RTH1/FEN1 nuclease recognition of the substrate. The presence of CDDP at the 5Ј-end did not affect RTH1/FEN1 nucleolytic cleavage, as can be seen by the appear-  Fig. 1. The downstream primer sequence is 5Ј-CCTATGGCCAATAACCTTATCTC-TACTCCAATCAATTTAACTTCCACTTTCCTCACCCCA-3Ј. The tail is underlined, and the platinated residues are in bold. The template sequence is 5Ј-CTGGGGTGAGGAAAGTGGAAGTTAACCACGGTCGTG-AATGGGTTTTTTTT-3Ј. B, unmodified control DNA substrate was used in lanes 1-6, whereas lanes 7-12 have a DNA substrate with a modification of the sugar:phosphate backbone between nt 6 and 7 from the 5Ј-end of a 50-nt tail. The modified nucleotide is shown in bold. Reactions were carried out with hRTH1 nuclease for the length of time indicated. The upstream primer and template sequences are the same as those used in Fig. 3. The downstream primer sequence is 5Ј-CCGA-GCTCGAATTCGCCCGTTTCACGCCTGTTACTTAATTCACTGGCCG-TCGTTTTACAACGACGTGACTGG-3Ј. The tail is underlined. ance of cleavage products over time, corresponding to the short 8-nt tail (Fig. 5).
CDDP Adducts Near the Cleavage Site Are Inhibitory-Although RTH1/FEN1 nuclease can traverse CDDP adducts, we considered that the adduct could still have a direct inhibitory effect on the catalysis of cleavage. As mentioned earlier, the binding of CDDP to double stranded DNA creates a helical distortion that disrupts base pairing 1 or more nucleotides on either side of the binding site. It might also distort necessary interactions of amino acid residues within the active site of RTH1/FEN1 with the substrate. We positioned CDDP at the first 2 unannealed nucleotides of the 5Ј-tail to determine whether this would interfere with RTH1/FEN1-directed cleavage. In this situation, cleavage was inhibited (Fig. 6). DISCUSSION The mechanism of endonucleolytic cleavage by the RTH1/ FEN1 nuclease involves sliding over the 5Ј-end of the unannealed tail of a primer annealed to a template, traversing the unannealed region, and cleaving near the first annealed nucleotide (18). Results presented here show that the mammalian nuclease can slide over a variety of covalent adducts to either the bases or the sugar-phosphate backbone of DNA. This specificity is consistent with genetic evidence in yeast that absence of the RTH1/FEN1 nuclease results in a defect in adduct repair (12).
In view of the tracking requirements of RTH1/FEN1 nuclease, we considered that it could slide over the tail by one of two mechanisms. In the first, the nuclease would fully encircle the unannealed tail. In this case, it would be analogous to proliferating cell nuclear antigen, which encircles and slides freely on double stranded DNA (40). The second involves passing the DNA or RNA tail through a groove created by the tertiary structure of the protein. In the first case, there should be a limit on adduct size, beyond which the passage of the nuclease would be blocked. However, if the nuclease tracked by means of a groove, tolerance of adducts could be more complex. For example, if tracking involved protein-DNA contacts with the sugarphosphate backbone, as opposed to contact with the bases, there may be no limit on the size of base adducts. However, even very small adducts on the backbone would block tracking. Remarkably, our results show that both CDDP adducts to the bases and tert-butyl-silyl group additions to the sugars fail to block tracking. These observations clearly show that the tracking process does not require protein-DNA contacts that depend on normal base and sugar structures.
Recently the crystal structures have been determined for T5 5Ј-exonuclease (41), the Thermus aquaticus DNA polymerase 5Ј-nuclease (42), and the T4 RNase H (43), which are functionally homologous to the mammalian 5Ј-to 3Ј exonuclease/endonuclease. These prokaryotic nucleases both have a distinctive "helical arch" structure. Ceska et al. (41) have proposed that the unannealed tail is threaded through the fully encircled hole in the protein created by the arch. Interestingly, the hole can clear single but not double stranded DNA. Furthermore, the arch appears to be sufficiently flexible that one can envision passage of adducts on the single strand. It is possible that the mammalian nuclease has a similar structure and that the  1-5) contain nonplatinated DNA; lanes 6 -10 contain DNA substrates with a CDDP adduct on the tail of a downstream primer, adjacent to the cleavage junction as depicted above the autoradiographed gel. Reactions were carried out for the length of time indicated. Two bands are seen in the control lanes, because the downstream primer has some DNA that was not synthesized to full length. The upstream primer sequence is 5Ј-ATATGGCTC-GATGCCGATCC-3Ј. The downstream primer sequence is 5Ј-CACCC-CACTCCTTTCACCTTTCCCCGGTCCCACCACTTATCCATTTAAAA-ACCAAAAAAAAAAA-3Ј. Platinated nucleotides are in bold, and the tail is underlined. The template sequence is 5Ј-GTTTTTTTTTTTGG-TTTTTAAATGGATAAGTGGTGGGAGGATCGGCATCGAGCCAT-AT-3Ј. bubble substrate cannot be cleaved, because passage of a 5Ј-tail though the arch is a prerequisite for correct interaction with the cleavage site.
The proposed participation of an arch anticipates an ultimate limit on acceptable adduct size. The large platinum adduct we have used not only covalently attaches adjacent guanosine residues but also has bulky side chains, producing a structure very different from a normal dinucleotide. The RTH1/ FEN1 nuclease efficiently passed this structure, suggesting that if the nuclease encircles the strand, the passage through the protein is designed to accommodate large adducts. We never saw an adduct size limit cleavage of RTH1 in current experiments. However, previous studies determined that although biotinylated nucleotides would allow passage of the nuclease, such nucleotides bound to streptavidin would not. Also, single stranded binding protein bound to the strand is large enough to block tracking. This shows that there is a maximum limit on adduct size. However, the proteins are so large they could block tracking by either of the proposed mechanisms.
Footprint analysis provided additional information about how the nuclease could be contacting the tracking strand. A MNase-generated footprint of the nuclease, in the absence of magnesium, presumably shows that it bound in the position it assumes just before catalysis of cleavage. At that time, approximately 25 nucleotides of the tail are protected adjacent to the cleavage junction. This is a long footprint for an enzyme with a molecular mass of 44 kDa, but not unusual. 2 Size analysis of the purified nuclease suggests that it is a monomer (3). The free sliding mechanism of interaction with the substrate does not suggest that there is a multimerization during sliding and catalysis. Possibly the nuclease is long and thin. Alternatively, the tail may be wound around the nuclease for a distance of 25 nucleotides.
Considering the large footprint, the part of the nuclease responsible for tracking could reside in the region that contacts the tail, for example, 20 -25 nucleotides away from the annealing point. In this situation, the nuclease could have been insensitive to adducts near the point of cleavage but inhibited by adducts 20 or more nucleotides toward the 5Ј-end of the tail. To examine this possibility, we generated substrates in which the damaged site was beyond the footprinted region, obligating the nuclease to fully traverse the damage prior to cleavage. A CDDP or a tert-butyl-silyl group was placed 30 or 44 nucleotides, respectively, from the cleavage site. Neither substrate was inhibitory to RTH1, indicating that there is enough spatial clearance throughout the DNA binding region to completely bypass these adducts.
In view of the high tolerance of the RTH1/FEN1 nuclease for a variety of adduct structures, we considered the possibility that adduct damage produces a conformation in the substrate that suspends the 5Ј-end and tracking requirements. In this case, adducts are tolerated because they are never encountered. To test this possibility, we used the CDDP adduct to produce a bubble, i.e. a melted region in a double stranded DNA. This resembles the RTH1/FEN1 substrate but lacks the free 5Ј-end. However, the damage-generated bubble is inert to cleavage by RTH1 nuclease, as are bubbles produced by mispairing (8). Therefore, the substrate for damage removal is most likely produced by a damage-specific nuclease that cleaves 5Ј of the adduct or by a DNA replication-derived break in the chromosome.
We have demonstrated the ability of RTH1/FEN1 nuclease to bypass CDDP adducts during tracking on an unannealed tail. However, not all positions on the tail resist the inhibitory effect of a cis-platinum adduct. As seen in Fig. 6, cleavage is inhibited if the adduct is on the first 2 nucleotides adjacent to the annealing point. This effect does not appear to result from inhibition of tracking. A likely reason is that CDDP at this position interferes with the actual catalysis of cleavage by distorting interactions that the nuclease needs to make with the nucleotides in the tail adjacent to the cleavage site. Another possibility is that the adduct prevents proper helix structure in the adjacent double stranded region. This would create a longer tail and a gap between the upstream primer and the cleavage site. Since the nuclease prefers an adjacent upstream primer on this substrate, the production of a gap could be inhibitory. In a repair reaction in vivo, this situation would not be a problem, since synthesis from the upstream primer could displace more of the tail and move the adduct to a position farther out onto the single stranded region.
We have previously reconstituted Okazaki fragment processing in vitro and demonstrated that RTH1/FEN1 nuclease can participate in removal of the initiator RNA (2). More recently we demonstrated that one possible pathway for this process is displacement of the 5Ј-end of the RNA-initiated Okazaki fragment, followed by tracking and endonucleolytic cleavage by the nuclease in the region of the DNA just beyond the initiator RNA (44). In this way the nuclease could operate by the same mechanism for both replication and repair. We found that cRTH1/FEN1 nuclease co-purifies with DNA polymerase ⑀ (3) and a 3Ј-to 5Ј-DNA helicase, with respect to template (45), through a number of chromatographic steps. The polymerase is required for cellular viability (46) and has been postulated to participate in lagging strand DNA replication and essential DNA repair functions (47)(48)(49). Additionally, recent work indicates that S. cerevisiae dna2 3Ј-to 5Ј-helicase interacts with yeast RTH1/FEN1 nuclease in the two-hybrid system (50,51). The minimal requirements for repair of adduct damage after incision could be met by polymerase, 3Ј-to 5Ј-helicase, and RTH1/FEN1 functions. A model of these enzymes interacting 2 J. Hayes, personal communication. with a damaged DNA substrate is shown in Fig. 7. Following incision 5Ј to the site of damage, a 3Ј-to 5Ј-helicase would displace the damaged section of DNA. RTH1/FEN1 nuclease could endonucleolytically remove this damaged 5Ј-tail, as a polymerase, such as DNA polymerase ⑀, fills in the gap. Finally, the nick could be sealed by DNA ligase I.
These data show the potential for RTH1/FEN1 nuclease to participate in some types of adduct repair. Although removal of the CDDP adduct has been shown to be perfomed by nucleotide excision repair (25)(26)(27), the exact enzymes involved in different types of adduct repair have not all been elucidated. Genetic evidence indicates that RTH1/FEN1 nuclease is necessary for efficient repair of methyl methane sulfonate, an alkylating agent (12). We suggest that other types of adduct damage could be removed by the action of RTH1/FEN1 nuclease through a mechanism alternative to nucleotide excision repair.