The Exonucleolytic and Endonucleolytic Cleavage Activities of Human Exonuclease 1 Are Stimulated by an Interaction with the Carboxyl-terminal Region of the Werner Syndrome Protein*

Exonuclease 1 (EXO-1), a member of the RAD2 family of nucleases, has recently been proposed to function in the genetic pathways of DNA recombination, repair, and replication which are important for genome integrity. Although the role of EXO-1 is not well understood, its 5′ to 3′-exonuclease and flap endonuclease activities may cleave intermediates that arise during DNA metabolism. In this study, we provide evidence that the Werner syndrome protein (WRN) physically interacts with human EXO-1 and dramatically stimulates both the exonucleolytic and endonucleolytic incision functions of EXO-1. The functional interaction between WRN and EXO-1 is mediated by a protein domain of WRN which interacts with flap endonuclease 1 (FEN-1). Thus, the genomic instability observed in WRN–/– cells may be at least partially attributed to the lack of interactions between the WRN protein and human nucleases including EXO-1.

Recently we reported a novel interaction of WRN protein with the human 5Ј-flap endonuclease/5Ј to 3Ј-exonuclease (FEN-1) (24), a DNA structure-specific nuclease implicated in DNA replication, recombination, and repair (25). WRN protein stimulates FEN-1 cleavage activity by a physical interaction with a COOH-terminal domain of the WRN protein (24). Another member of the RAD2 family to which FEN-1 belongs is human exonuclease 1 (EXO-1) (26). Like FEN-1, EXO-1 is a structure-specific endonuclease as well as an exonuclease (27,28). EXO-1 has been implicated in a number of DNA metabolic pathways including mismatch correction, mitotic and meiotic recombination, and double strand break repair (29 -34). A role for EXO-1 in Okazaki fragment processing during DNA replication has been suggested based on its structure-specific endonuclease and RNase H activities that are similar to those of FEN-1 (30). Functional overlap between EXO-1 and FEN-1 (Rad27 in Saccharomyces cerevisiae) has been proposed from observations in yeast that exoI: rad27 double mutants are inviable (35) and that overexpression of yeast EXO-1 or human EXO-1 complements cellular phenotypes of rad27 mutants (30,35,36).
The evident sequence homology and similar biochemical activities of EXO-1 and FEN-1, as well as the potential functional redundancy of the two nucleases, suggested to us that WRN might also interact functionally with EXO-1. We have found this to be the case. Evidence is presented that WRN interacts physically with human EXO-1 and stimulates the endonucleolytic and exonucleolytic cleavage activities of EXO-1. The functional interaction is independent of WRN catalytic function and mediated by a COOH-terminal region of WRN which also interacts with FEN-1. Thus, WRN modulates the cleavage activities of both human EXO-1 and FEN-1 by a direct protein interaction, suggesting that either structure-specific nuclease may act together with WRN during DNA replication, recombination, or repair. The physical and functional interaction of WRN with these human nucleases is likely to be important for the cellular role of WRN in the maintenance of genome integrity.

MATERIALS AND METHODS
Proteins-Hexahistidine-tagged recombinant human EXO-1 protein was overexpressed using a baculovirus/insect system and purified as described elsewhere. 2 The purified EXO-1 protein was judged to be Ͼ97% pure from analysis on Coomassie-stained SDS-polyacrylamide gels (see Fig. 2A). Baculovirus constructs for full-length recombinant * 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.
hexahistidine-tagged WRN proteins (wild-type, WRN-K577M, WRN-E84A) or a truncated version of WRN containing only the aminoterminal 368 amino acids of the protein (designated N-WRN 1-368 ) were kindly provided by Drs. M. Gray (University of Washington, Seattle) and J. Campisi (Lawrence Berkeley National Laboratory, Berkeley, CA). Amplified WRN-encoding baculovirus was used to infect Sf9 cells for overexpression of WRN protein as described elsewhere (37). A recombinant hexahistidine-tagged carboxyl-terminal fragment of WRN (residues 940 -1432, designated C-WRN 940 -1432 ) was overexpressed in Escherichia coli and purified as described previously (38). Recombinant human FEN-1 was purified as described previously (24). Human PCNA and RPA were graciously provided by Dr. M. Kenny (Albert Einstein College of Medicine, Bronx, NY).
WRN-EXO-1 Coimmunoprecipitation Experiments-HeLa cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum at 37°C in 5% CO 2 . Nuclear extract was prepared as described previously (39). For coimmunoprecipitation, HeLa nuclear extract (1.5 mg of protein) was incubated with goat anti-WRN polyclonal antibody (1:40; Santa Cruz Biotechnology) in buffer D (50 mM HEPES pH 7.5, 100 mM KCl, 10% glycerol) for 4 h at 4°C and tumbled with 20 l of protein G-agarose (Roche Applied Science) at 4°C overnight. Beads were washed three times with buffer D supplemented with 0.1% Tween 20. Proteins were eluted by boiling treatment in SDS sample buffer, resolved on 8% polyacrylamide Tris-glycine SDS gels, and transferred to PVDF membranes (Amersham Biosciences). The membranes were blocked with 5% nonfat dry milk in phosphate-buffered saline containing 0.1% Tween 20 and probed using rabbit polyclonal antibody against a recombinant human EXO-1 nuclease domain fragment characterized previously (27) (1:5,000; courtesy of Dr. D. M. Wilson III (NIA, NIH)) followed by goat anti-rabbit IgG conjugated to horseradish peroxidase (Santa Cruz Biotechnology) or probed for WRN using mouse anti-WRN monoclonal antibody (1:250; BD Pharmingen) followed by goat-anti-mouse IgG-horseradish peroxidase (Vector Laboratories). EXO-1 or WRN on immunoblot was detected using ECL Plus (Amersham Biosciences).
Coimmunoprecipitation of purified WRN and EXO-1 was performed in the presence of binding buffer (50 mM Tris, pH 8.0, 10% glycerol, 100 mM NaCl, 0.01% Nonidet P-40) containing protease inhibitors (0.1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, and 5 g/ml leupeptin). In a 100-l reaction volume, 500 ng of purified WRN and 500 ng of EXO-1 were incubated with goat anti-WRN polyclonal antibody (1:40) in binding buffer for 4 h at 4°C. The protein complex was adsorbed on to protein G-agarose beads by incubating the mixture overnight at 4°C with gentle rotation. The beads were washed three times with binding buffer, eluted by boiling with SDS sample buffer, and resolved on 8 -16% gradient SDS-polyacrylamide Tris-glycine gels. After transferring the proteins to PVDF membranes, the membranes were blocked with 5% nonfat dry milk in phosphate-buffered saline containing 0.1% Tween 20 and probed with either anti-EXO-1 or anti-WRN antibodies as described above.
Far Western Blotting-Far Western blotting was conducted as described previously (40). Briefly, 0.2-1.0 g of each protein was electrophoresed on 8 -16% SDS-polyacrylamide gels, transferred to PVDF membranes, and processed as described previously (40) with the exception that specified blocked membranes were incubated in the presence of 0.5 g/ml EXO-1. Membranes were washed, and conventional West-ern analysis was then performed to detect the presence of EXO-1 using rabbit anti-EXO-1 polyclonal antibody (described above) (1:5,000). Goat anti-rabbit IgG-horseradish peroxidase conjugate was used as secondary antibody at a 1:10,000 dilution and detected using ECL Plus.
Oligonucleotide Substrates-PAGE-purified oligonucleotides (Midland Certified Reagent Co.) (Table I) were used for preparation of substrates. Substrates were prepared as described previously (41). Briefly, 10 pmol of the appropriate flap oligonucleotide was 5Ј-radiolabeled with [␥-32 P]ATP and T4 polynucleotide kinase (New England Biolabs) using the manufacturer's protocol. Unincorporated nucleotide was removed using a Sephadex G-25 spin column (Amersham Biosciences). The radiolabeled oligonucleotide was annealed to 25 pmol of the appropriate template oligonucleotide by heating at 95°C, then cooling down from 70 to 24°C. If necessary, 50 pmol of an upstream oligonucleotide was then annealed to the duplex substrate by heating at 37°C for 1 h and slowly cooling to 24°C. Two nicked duplex substrates were prepared: 1) nicked duplex A using FLAP00A, TSTEM, and U25; and 2) nicked duplex B using FLAP00B, TSTEM2, and U21. The monodeoxynucleotide (nt) flap was prepared with FLAP01, TSTEM, and U25. For the monoribodeoxynucleotide flap, FLAP01RNA, TSTEM, and U25 were used. The 10-nt flap was created with FLAP10, TSTEM2, and U21. The blunt ended duplex substrate was made using TSTEM-COMP and TSTEM.
For the nicked duplex substrate with a 3Ј-end label on the upstream primer, 10 pmol of U24 was annealed to 25 pmol of TSTEM and end labeled with [␣-32 P]dCTP and Klenow fragment (New England Biolabs) at 25°C for 20 min followed by an additional incubation for 20 min at 25°C with 50 M unlabeled dCTP. Unincorporated nucleotide was removed by passing over two Sephadex G-25 columns. Klenow was then heat inactivated at 95°C followed by slow cooling to permit the oligonucleotides to reanneal. 50 pmol of FLAP00 was then added and annealed by heating at 37°C for 1 h followed by slow cooling.
EXO-1 Incision Assay-20-l reactions contained 0.5 nM DNA substrate (unless otherwise noted) and the indicated concentrations of WRN and/or EXO-1 in 30 mM HEPES pH 7.6, 5% glycerol, 40 mM KCl, 0.1 mg/ml bovine serum albumin, and 8 mM MgCl 2 . WRN was mixed with the substrate on ice prior to the addition of EXO-1. Reactions were incubated at 37°C for 15 min (unless indicated otherwise), terminated with the addition of 10 l of formamide dye (80% formamide (v/v), 0.1% bromphenol blue, and 0.1% xylene cyanol), and heated to 95°C for 5 min. Products were resolved on 20% polyacrylamide, 7 M urea denaturing gels. A PhosphorImager was used for detection, and the Image-Quant software (Molecular Dynamics) was used for quantitation of the reaction products. The percent incision was calculated as described previously (24).
WRN Helicase Assay-20-l reactions contained 0.5 nM 10-nt 5Ј-flap DNA substrate (except where indicated) and the indicated concentrations of WRN and/or EXO-1 in the same reaction buffer as used for EXO-1 incision assays except for the additional presence of 2 mM ATP. Reactions were incubated at 37°C for 15 min, terminated with the addition of 20 l of helicase stop solution (50 mM EDTA, 40% glycerol, 0.1% bromphenol blue, 0.1% xylene cyanol) containing a 10-fold excess of unlabeled oligonucleotide with the same sequence as the labeled strand. The products of the helicase reactions were resolved on nondenaturing 12% polyacrylamide gels. Radiolabeled DNA species in polyacrylamide gels were visualized using a PhosphorImager and quanti- 25 CGCCAGGGTTTTCCCAGTCACGACC tated using the ImageQuant software. The percent helicase substrate unwound was calculated as described previously (42).

Physical Interaction between WRN and EXO-1-
The previously reported physical interaction between WRN and FEN-1 (24) suggested to us that EXO-1, a RAD2 family member and structure-specific nuclease like FEN-1 (27,28), might interact directly with WRN. To explore this possibility, we tested for the coimmunoprecipitation of WRN and EXO-1 from human nuclear extracts (Fig. 1). Using 200 g of protein from HeLa nuclear extract, we detected EXO-1 (Fig. 1, lane 1) that comigrated with the purified recombinant human EXO-1 used in this study (Fig. 1, lane 4) by Western blot analysis. An antibody directed against WRN protein reproducibly immunoprecipitated EXO-1 from 1.5 mg of HeLa nuclear extract as detected by Western blot with an anti-EXO-1 antibody (Fig. 1, lane 2). The immunoprecipitation of WRN from HeLa nuclear extract with anti-WRN antibody was verified by Western blot analysis using anti-WRN antibody (data not shown). Preimmune IgG failed to precipitate either EXO-1 (lane 3) or WRN (data not shown) in control experiments. These results provide evidence that WRN and EXO-1 are associated with each other in human nuclei, suggesting that the two proteins may physically interact with each other.
To explore the specificity and nature of the suggested WRN-EXO-1 interaction, coimmunoprecipitation experiments were performed with the purified recombinant proteins. Using an antibody directed against WRN, both WRN and EXO-1 were immunoprecipitated ( Fig. 1, B and C, lane 5). WRN and EXO-1 were also coimmunoprecipitated in the presence of ethidium bromide confirming that the protein interaction is not mediated through DNA (Fig. 1, B and C, lane 6). Approximately 20% of the input EXO-1 protein was precipitated by the WRN antibody. The purified proteins failed to be immunoprecipitated by goat preimmune serum (Fig. 1, B and C, lane 4) or in the absence of antibody (Fig. 1, B and C, lane 2). In control experiments, EXO-1 failed to be immunoprecipitated by anti-WRN antibody when WRN protein was omitted from the binding mixture ( Fig. 1, B and C, lane 3), attesting to the specificity of the WRN antibody used for the coimmunoprecipitation studies. These results indicate a direct physical interaction between WRN and EXO-1 and support the evidence that WRN and EXO-1 interact with each other in vivo.
We next sought to determine whether the COOH-terminal region of WRN responsible for the physical and functional interaction with FEN-1 also interacted physically with EXO-1. To test for this interaction, we performed far Western blotting experiments. For these studies, we tested the ability of fulllength WRN protein, a COOH-terminal fragment of WRN (C-WRN 940 -1432 ), and an NH 2 -terminal fragment of WRN (N-WRN 1-368 ) to bind EXO-1. These proteins are shown in a Coomassie-stained SDS-polyacrylamide gel in Fig. 2A. The proteins were resolved on SDS-polyacrylamide gels and transferred to PVDF membranes that were subsequently blocked, incubated in buffer alone or buffer containing EXO-1, and washed extensively. The blots were then probed with an antibody against EXO-1. As shown in Fig. 2B, EXO-1 bound to full-length WRN and the COOH-terminal fragment of WRN (lanes 1 and 2) but did not bind to the NH 2 -terminal fragment of WRN (lane 3). In the control experiment, when a similar membrane was incubated with buffer lacking EXO-1 and probed with anti-EXO-1 antibody, no bands corresponding to the position of full-length WRN, COOH-terminal or NH 2 -terminal WRN were detected (Fig. 2C, lanes 1, 2, and 3). In control experiments, two faint bands (fast and slow migrating) were detected in lane 4 containing bovine serum albumin, but this signal was caused by nonspecific reactivity of the EXO-1 antibody, as these bands were also detected on membranes that were incubated in the presence or absence of EXO-1 protein.
These results together with the immunoprecipitation experiments demonstrate that WRN and EXO-1 interact physically and that the COOH-terminal region of the WRN protein alone is capable of binding to EXO-1.
WRN Stimulates EXO-1 Cleavage of a 1-Nucleotide 5Ј-Flap Substrate-The physical interaction between WRN and EXO-1 and the ability of WRN to stimulate FEN-1 cleavage by a protein interaction (24) suggested that WRN might stimulate the incision activities of human EXO-1. Under the same reaction conditions, a 1-nt 5Ј-flap substrate was susceptible to EXO-1 cleavage that generated the 1-nt product (Fig. 3A, lane 2), whereas FEN-1 cleavage of the same substrate resulted primarily in the 2 nt product, and to a lesser extent the 1-nt product (Fig. 3A, lane 6), consistent with previous observations (24). In the presence of WRN (4 nM), both EXO-1 and FEN-1 cleavage reactions were stimulated to yield the same respective cleavage products (Fig. 3A, lanes 3 and 7). Using a limiting concentration of EXO-1 (0.125 nM), the cleavage reaction was stimulated 8-fold by 8 nM WRN (Fig. 3B, lanes 5 and 6, and 3C). Using a 2-fold higher concentration of EXO-1 (0.25 nM), 10% of the DNA substrate was cleaved (Fig. 3B, lane 8, and 3C). At this EXO-1 level, WRN stimulated EXO-1 cleavage to 43% of the substrate incised (Fig. 3B, lane 9, and 3C). In comparison, 0.25 nM EXO-1 did not appreciably incise the 5Ј-32 P end-labeled blunt duplex 44-bp substrate, and the presence of 8 nM WRN resulted in only a small stimulation of incision activity (ϳ1.5%) using 0.25 nM EXO-1; however, at a higher concentration of EXO-1 (1 nM), the blunt duplex DNA substrate was degraded from its 5Ј-end, and the EXO-1 reaction was stimulated by WRN in a dose-dependent manner (data not shown). These results indicate that EXO-1 incision at the internally positioned 1-nt flap was primarily responsible for the removal of the unpaired nucleotide rather than 5Ј to 3Ј-exonuclease digestion from the end of the blunt duplex substrate. WRN stimulation of EXO-1 cleavage of the 1-nt flap substrate was also observed in reactions containing 0.5 nM EXO-1 (Fig. 3B, lanes  10 and 11), although the level of stimulation was not as great because a plateau of incision activity (55%) was approached in the reactions containing EXO-1 and WRN (Fig. 3C). 8 nM WRN alone did not catalyze significant cleavage of the 1-nt flap DNA substrate (Fig. 3B, lane 12), consistent with previous observations (24). We also observed a significant stimulation of EXO-1 cleavage of a single ribonucleotide 5Ј-flap substrate in the presence of WRN using a limiting amount of EXO-1 (0.125 nM). 3 We subsequently studied EXO-1 cleavage as a function of WRN concentration. A limiting amount of EXO-1 (0.125 nM) was used such that cleavage of the 1-nt flap substrate was very low (ϳ3%) (Fig. 3D). A 3-fold stimulation of EXO-1 cleavage was detected at a WRN concentration of 2 nM (Fig. 3D). In the presence of 4 nM WRN, EXO-1 cleavage increased to 22% incision (Fig. 3D), a 7-fold stimulation of EXO-1 cleavage in the 15-min incubation. At 8 nM WRN, product formation began to plateau at ϳ30% (Fig. 3D).
Kinetic analysis of the EXO-1 cleavage reaction on the 1-nt 5Ј-flap DNA substrate demonstrated that the presence of WRN profoundly affected the rate of EXO-1 incision (Fig. 4). In these studies, we used a concentration of WRN (8 nM) which was previously determined to achieve maximal stimulation of EXO-1 cleavage (Fig. 3D). The concentration of EXO-1 (0.125 nM) used resulted in a low, but reproducibly detectable incision of ϳ2% of the 0.5 nM DNA substrate in a 15-min reaction in the absence of WRN (Fig. 3D). Stimulation of EXO-1 incision by WRN was detected at time points as short as 1-2 min (Fig. 4). Up to 9 min, EXO-1 cleavage in the absence of WRN was ϳ1%; however, in the presence of WRN, EXO-1 incised 19% of the DNA substrate at the 9 min time point. Regression analysis of the linear regions of the slopes yielded reaction rates of 25.9 and 0.7 pmol of product/min for the WRN ϩ EXO-1 and EXO-1 reactions, respectively. This difference represented a 37-fold increase in the rate of EXO-1 cleavage when WRN is present. At 15 min, the EXO-1 cleavage reaction conducted in the presence of WRN achieved a plateau of ϳ28% substrate incised. In contrast, EXO-1 alone cleaved only 1.5% of the substrate by the end of 15 min. These results demonstrate conclusively that WRN stimulates the rate of EXO-1 cleavage.
Catalytic Activities of WRN Are Not Required for Stimulation of EXO-1 Incision-Our previous work demonstrated that a protein domain of WRN, devoid of catalytic activities and which interacts physically with FEN-1, mediated the functional interaction between WRN and FEN-1 (24). However, it was conceivable that WRN might affect the cleavage activity of EXO-1 by a mechanism distinct from that of the WRN-FEN-1 interaction. The three catalytic activities of WRN (ATPase, helicase, exonuclease) might influence the functional interaction with EXO-1. To address this possibility, we tested the effects of full-length recombinant WRN proteins with site-directed mutations in the active sites of its catalytic domains on EXO-1 cleavage. WRN-K577M mutant protein, devoid of ATPase or helicase activity (18,43), was capable of stimulating EXO-1 cleavage of the 1-nt flap DNA substrate similarly to wild-type WRN resulting in ϳ30% incision (Fig. 5A, lanes 3 and 6) compared with 2% incision in reactions containing only EXO-1 (Fig.  5A, lane 2). WRN-E84A, defective in exonuclease activity (21,38), was able to stimulate EXO-1 cleavage similarly to the wild-type or ATPase mutant WRN proteins (Fig. 5A, lane 4). The WRN-K577M mutant protein preparation contained a small level of nuclease activity which digested ϳ2% of the substrate (Fig. 5A, lane 10), whereas the WRN-E84A was free of nuclease activity on the 5Ј-flap substrate (Fig. 5A, lane 8).
These results indicate that WRN exonuclease or helicase/ ATPase activities are not required for the stimulation of EXO-1 nuclease activity by WRN. We also tested a truncated recombinant WRN protein (residues 1-368) which lacks the ATPase/ helicase and COOH-terminal region of full-length WRN protein but retains the exonuclease domain and possesses intrinsic 3Ј to 5Ј-exonuclease activity (44). The N-WRN 1-368 protein failed to stimulate EXO-1 incision (Fig. 5A, lane 5), suggesting that the functional interaction between WRN and EXO-1 is specific and may be mediated by the COOH-terminal region of WRN.
The ability of the catalytically defective mutant WRN proteins to stimulate EXO-1 cleavage suggested that the functional interaction is mediated by a direct protein interaction. We demonstrated previously that a COOH-terminal fragment (sequence 940 -1432) of WRN which physically interacts with FEN-1 is capable of stimulating FEN-1 cleavage of a 1-nt 5Ј-flap substrate (24). To determine whether this domain of WRN is sufficient to stimulate EXO-1 cleavage, we tested directly the effect of purified C-WRN 940 -1432 on EXO-1 cleavage of the 1-nt 5Ј-flap substrate (Fig. 5B). 0.25 nM EXO-1 incised 1, 4, and 13% of the 1-nt 5Ј-flap substrate in the presence of 0, 4, and 8 nM C-WRN 940 -1432 , respectively (Fig. 5B, lanes 2, 3, and  4). No incision was detected in reactions containing only C-WRN 940 -1432 (Fig. 5B, lanes 5 and 6). The ability of the highly purified C-WRN 940 -1432 fragment to stimulate EXO-1 cleavage indicates that the protein domain responsible for stimulating EXO-1 cleavage resides within the COOH terminus of WRN, a region of the protein which interacts physically with EXO-1 and was shown to mediate the functional interaction with another Rad2 family member, human FEN-1 (24).
WRN Stimulates EXO-1 Incision of a Nicked Duplex DNA Substrate-It was observed recently that EXO-1 is required in a reconstituted mismatch repair reaction using human nuclear extracts with a DNA substrate where mismatch provoked incision is directed by a strand break located either 5Ј or 3Ј to the mismatch (45). A reported mismatch repair deficiency in WS fibroblast cell lines may be attributed to either a direct role of WRN in the repair pathway or the consequence of a secondary mutation in a mismatch repair gene which results in a loss of mismatch repair activity (46). We therefore set out to test the effect of WRN on EXO-1 cleavage activity of a proposed DNA mismatch repair intermediate, a nicked duplex DNA substrate. Two nicked duplex substrates of different sequences were tested with the 5Ј-32 P label residing on the 19-nt and 21-nt downstream primers of nicked duplex substrates A and B, respectively. The nicked duplex substrates were verified for their integrity by performing ligation reactions using T4 DNA ligase. Approximately 95% of nicked duplex substrates was converted to a ligated product of the upstream and downstream primers, providing evidence that they are indeed nicked duplex molecules (data not shown). 8 nM WRN stimulated 0.25 nM EXO-1 to remove the 5Ј-32 P-labeled nt of the downstream primer of either nicked duplex substrate A or B (Fig. 6A, lanes  3 and 7). In control reactions, 8 nM WRN alone did not catalyze any significant incision of the nicked duplex substrate, whereas higher concentrations of EXO-1 cleaved nicked duplex substrate A or substrate B (data not shown).
A quantitative analysis of EXO-1 cleavage of nicked duplex substrate A in the presence of 8 nM WRN as a function of EXO-1 concentration is shown in Fig. 6B. 0.5 nM EXO-1 incised 12% of nicked duplex substrate A. Using a limiting concentration of EXO-1 (0.125 nM), the cleavage reaction was stimulated 8-fold by 8 nM WRN, resulting in 26% incision (Fig. 6B). Using a 2-fold higher concentration of EXO-1 (0.25 nM), 5% of the nicked DNA substrate was incised (Fig. 6B). At this EXO-1 level, WRN stimulated EXO-1 incision to 42% of the substrate acted upon. Similar results were obtained using 0.5 nM EXO-1, although the degree of stimulation (ϳ6-fold) was not as great because the incision level began to plateau (Fig. 6B).
We next examined EXO-1 cleavage of the nicked duplex DNA substrate A as a function of WRN concentration. A limiting concentration of EXO-1 (0.125 nM) was used such that incision of the nicked duplex was very low (ϳ1.5%) (Fig. 6C). A statistically significant stimulation of EXO-1 cleavage was observed at a WRN concentration of 2 nM (Fig. 6C). At this WRN level, EXO-1 incised ϳ20% of the nicked DNA substrate (Fig. 6C). Further increase of the WRN concentration resulted in greater amount of EXO-1 digestion of the nicked duplex. At 20 nM WRN, EXO-1 cleaved 33% of the substrate, a 15-fold increase compared with reactions containing only EXO-1 (Fig. 6C).
It was speculated that EXO-1 might play a role in 3Ј-heteroduplex repair by having a regulatory/structural role in 3Јexcision complex assembly or by a cryptic 3Ј to 5Ј hydrolytic activity (45). We tested a nicked duplex substrate with a 3Ј-end label on the upstream primer and found that 0.125 nM EXO-1 in the absence or presence of 20 nM WRN did not incise the 3Ј-label under conditions where the 5Ј-end label of the downstream primer was catalytically removed (data not shown), suggesting that WRN does not reverse the 5Ј to 3Ј polarity of the exonuclease reaction catalyzed by EXO-1 at the site of a nick in a DNA duplex substrate. Furthermore, the results with the nicked duplex substrate containing a 3Ј-end label on the upstream primer suggest that the incision detected on the nicked duplex with the 5Ј-label on the downstream primer is caused by WRN stimulation of EXO-1 cleavage at the site of the nick. We conclude that the 5Ј to 3Ј-exonuclease activity of EXO-1 at a nick on duplex DNA is stimulated by WRN, and EXO-1 does not exhibit a 3Ј to 5Ј-exonuclease activity on these substrates under these conditions. WRN Stimulates the Endonucleolytic Cleavage Activity of EXO-1-Recent biochemical evidence has demonstrated that full-length S. cerevisiae EXO-1 as well as a protein fragment of human EXO-1 which harbors the conserved nuclease motifs of Rad2 family members (FEN-1, XPG, EXO-1) display 5Ј-flap endonuclease activity (27,28). It has been suggested that the structure-specific endonuclease activity of EXO-1 may be important for biological function in DNA metabolic pathways, most notably Okazaki fragment maturation during DNA replication. We therefore examined the effect of WRN on the flap endonuclease activity of EXO-1 to determine whether WRN might also stimulate structure-specific incision in addition to its positive effect on the 5Ј to 3Ј-exonuclease activity of EXO-1. The results shown in Fig. 7 demonstrate that this is the case. 1 nM EXO-1 endonucleolytically incised ϳ4% of a 10-nt 5Ј-flap substrate. The 10-nt incision product was the same as that shown previously to be endonucleolytically incised by a purified protein fragment of EXO-1 (27) (Fig. 7A, lane 2). Thus fulllength human EXO-1 has intrinsic flap endonuclease activity. We also observed some exonuclease activity on the 5Ј-flap as detected by the appearance of mononucleotide (Fig. 7A, lane 2), a result consistent with that observed when the purified nuclease domain fragment of EXO-1 was incubated with the same flap substrate (27). At low concentrations of WRN (0.5 and 1 nM), a small stimulation of EXO-1 incision (ϳ1.5-fold) was observed (Fig. 7A, lanes 4 and 5). Using 2 nM WRN, EXO-1 cleaved 9% of the 10-nt 5Ј-flap DNA substrate (Fig. 7A, lane 6). EXO-1 cleavage was elevated further in the presence of 4 or 8 nM WRN, yielding 21 and 31% incision, respectively (Fig. 7A,  lanes 7 and 8). Importantly, in control reactions 8 nM WRN alone did not incise the 10-nt 5Ј-flap structure (Fig. 7A, lane 9), as reported previously (42). At high concentrations of WRN (8 nM) (lane 9), a very low level of 5Ј-nuclease activity could be detected which was also evident in reactions containing exonuclease-defective WRN (data not shown), suggesting that a minor contaminant in the WRN protein preparation is responsible for this activity. Thus WRN is competent to stimulate the EXO-1 endonucleolytic cleavage reaction, as great as 8-fold, on the 10-nt 5Ј-flap substrate. We also observed stimulation of the EXO-1 5Ј to 3Ј-exonuclease activity on the 5Ј-ssDNA flap by WRN, resulting in 5 and 17% incision at WRN concentrations of 4 and 8 nM, respectively (Fig. 7A, lanes 7 and 8). The ability of WRN to stimulate 5Ј to 3Ј-exonuclease activity of EXO-1 on the 5Ј-ssDNA flap oligonucleotide is consistent with our observation that EXO-1 incision of 5Ј-32 P-labeled ssDNA oligonucleotide is also stimulated by WRN (data not shown).
We next tested the purified COOH-terminal protein fragment of WRN (C-WRN 940 -1432 ), devoid of the helicase and exonuclease domains, for its ability to stimulate EXO-1 cleavage of the 10-nt 5Ј-flap substrate. Using 1 nM EXO-1, the cleavage reaction was stimulated ϳ15-fold by 4 nM C-WRN 940 -1432 (Fig. 7B, lane 3). Slightly higher EXO-1 cleavage (37%) was observed using 8 nM C-WRN 940 -1432 (Fig. 7B,  lane 4). The 10-nt product from reactions containing C-WRN 940 -1432 and EXO-1 were also detected from reactions containing EXO-1 ϩ full-length WRN (Fig. 7A) or higher concentrations of EXO-1 (data not shown), indicating that the cleavage site specificity of EXO-1 is not altered by the WRN COOH-terminal protein fragment. Purified N-WRN protein fragment failed to stimulate endonucleolytic or exonucleolytic incision by EXO-1 (data not shown), adding further evidence that the functional interaction between WRN and EXO-1 is the result of a protein interaction mediated by the COOH-terminal region of WRN. The levels of stimulation of EXO-1 endonucleolytic cleavage by C-WRN 940 -1432 were comparable with those observed with the full-length WRN protein (Fig. 7B, lanes 7  and 8). These results indicate that the protein domain of WRN which physically binds to EXO-1 is capable of stimulating EXO-1 cleavage of the 10-nt 5Ј-flap substrate.
WRN Stimulates EXO-1 Cleavage as Effectively as FEN-1 Cleavage-Because WRN can stimulate either EXO-1 or FEN-1 endonucleolytic incision of 5Ј-flap substrates, we wanted to compare the -fold stimulation by WRN of the two human Rad2 nucleases. Using the 10-nt 5Ј-flap substrate, WRN was tested for its effect on EXO-1 or FEN-1 cleavage under the identical reaction conditions. The concentrations of EXO-1 (0.5 nM) and FEN-1 (0.06 nM) were adjusted to achieve a similar level of endonucleolytic cleavage (ϳ5%) of the 5Ј-flap substrate (Fig.  7C). In the presence of 8 nM WRN, EXO-1 cleavage was increased 6-fold, and FEN-1 cleavage was increased 7-fold (Fig.  7C). These results indicate that the stimulatory effect of WRN on flap cleavage by the two human endonucleases is very similar.
Both RPA and PCNA stimulate FEN-1 cleavage (41,47), raising the possibility that these human nuclear proteins might stimulate EXO-1 cleavage; however, there is no previous report of the effect of RPA or PCNA on human EXO-1 cleavage. At molar concentrations of PCNA homotrimer or RPA heterotrimer equal to WRN monomer (8 nM) that displayed a 16-fold stimulation of EXO-1 endonucleolytic cleavage of the 10-nt flap substrate, we did not detect any significant stimulation of EXO-1 cleavage (endonucleolytic or exonucleolytic) by either PCNA or RPA (data not shown).
WRN Stimulates EXO-1 Endonucleolytic Cleavage under Conditions Favorable for Helicase-catalyzed DNA Unwinding-Recently, we reported that WRN helicase efficiently unwinds a 10-nt 5Ј-flap substrate, releasing the 5Ј-flap oligonucleotide, leaving the duplex species containing the upstream oligonucleotide intact (42). To determine the outcome of 5Ј-flap processing by EXO-1 on a substrate that is also unwound by WRN, we examined both DNA unwinding (release of the 5Ј-flap oligonucleotide) and EXO-1 cleavage of the 5Ј-flap oligonucleotide in the presence of ATP, which fuels the WRN helicase reaction. The products were analyzed on nondenaturing polyacrylamide gels, and the results of a typical experiment are shown in Fig. 8. 4 or 8 nM WRN stimulated cleavage of the flap substrate by 0.5 nM EXO-1 ϳ10-fold and 6-fold, respectively (Fig. 8, lanes 3 and 4) as detected by the increased amount of the 10-nt incision product that comigrated with the radiolabeled 10-nt marker indicated on the left. Under these conditions in the presence of 0.5 nM EXO-1, 4 or 8 nM WRN unwound 8 and 39% of the flap substrate, respectively (Fig. 8, lanes 3 and  4). The oligonucleotide released by WRN helicase activity comigrated with the intact 31-mer (FLAP10) used in the preparation of the 5Ј-flap substrate. Similar levels of WRN helicase activity (14 and 44%) were detected in reactions containing only 4 or 8 nM WRN (lanes 5 and 6). When EXO-1 was increased to 2 nM, 7% of the 5Ј-flap substrate was endonucleolytically incised in the absence of WRN (lane 7). In the presence of WRN, endonucleolytic incision by EXO-1 was again increased significantly as detected by the appearance of the 10-nt incision product (lanes 8 and 9).

DISCUSSION
In this study we have demonstrated that WRN and EXO-1 interact physically as evidenced by their coimmunoprecipitation using either human extracts or purified recombinant proteins. The region of WRN which interacts physically with EXO-1 was mapped to the COOH-terminal region of WRN by far Western analysis. WRN robustly stimulates EXO-1 cleavage by a direct protein interaction that is independent of energy and catalytic activity. WRN exerts a marked enhancement (37-fold) in the rate of EXO-1 incision. Although a greater amount of WRN monomer compared with EXO-1 was necessary to achieve maximal stimulation of EXO-1 nuclease activity, the functional assembly state of WRN is not known and may assume an oligomeric state as proposed for some helicases. On a stoichiometric basis, WRN monomer is far superior to PCNA homotrimer or RPA heterotrimer in stimulating EXO-1 cleavage of short and long 5Ј-flap substrates. The results from these biochemical studies clearly indicate that WRN protein strongly stimulates the endonucleolytic and exonucleolytic incision activities of EXO-1 on proposed DNA intermediates of replication and repair. Importantly, WRN is the first protein identified that modulates the catalytic activities of EXO-1.
The ability of WRN to stimulate the incision reactions catalyzed by the sequence-related human nucleases EXO-1 and FEN-1 on short (1-nt) and long (10-nt) 5Ј-ssDNA flap substrates suggests that the functional interaction may proceed by a similar mechanism. In both cases, the stimulation of incision is mediated by a COOH-terminal domain of the WRN protein.
Although the WRN interaction site on FEN-1 is not known, it will be of interest to determine whether the site of WRN physical interaction is the same for EXO-1 and FEN-1. A number of similarities exist between EXO-1 and FEN-1, but there are some noticeable differences in their DNA substrate specificities and cleavage products (27,28). For the structures tested thus far, the DNA substrate specificities of EXO-1 and FEN-1 are not altered by the presence of WRN. WRN stimulates EXO-1 cleavage of ssDNA and blunt duplex dsDNA, whereas FEN-1 acts poorly on these substrates, and WRN does not appreciably stimulate FEN-1 on either of these structures (24). Although WRN is able to stimulate structure-specific endonucleolytic cleavage of 5Ј-flap structures by EXO-1 and FEN-1, the nuclease reactions display a subtle difference in the products generated. WRN stimulation of EXO-1 yields a 1-nt product from the 1-nt 5Ј-flap substrate, consistent with the cleavage products obtained from reactions containing higher levels of EXO-1 alone. Thus EXO-1 cleaves primarily at the base of the flap, yielding a nicked duplex intermediate that can be readily sealed by ligase. In contrast, FEN-1 (in the absence or presence of WRN) incises primarily 1 nt into the downstream annealed region, and to a lesser extent at the base of the flap, resulting in primarily a 1-nt gap duplex and a lower amount of nicked duplex, respectively (24,41). This difference may be important during Okazaki fragment processing of double flap substrates in which the upstream primer has a single unannealed nt, a preferred DNA substrate for FEN-1 cleavage (48). By the preferential FEN-1 incision of the 5Ј-flap 1 nt into the annealed downstream region, the unannealed (but complementary) nt at the 3Ј-end of the upstream primer may anneal to form a ligatable nicked DNA duplex substrate. It will be of interest to determine whether human EXO-1 also prefers a double flap structure as a cleavage substrate; alternatively, the cleavage specificity of EXO-1 may be well suited to process 5Ј-flap structures that lack 3Ј-flaps, a reaction that may also be important during eukaryotic DNA replication or some other pathway of DNA metabolism.
In which cellular pathways is the functional interaction of WRN and EXO-1 likely to be important? Yeast genetic studies have implicated EXO-1 in spontaneous mitotic and meiotic recombination between direct repeats (29). In one model, the 5Ј to 3Ј-exonuclease activity of EXO-1 may act to generate 3Ј-ssDNA tails that can be used by homologous pairing proteins for recombination. EXO-1 has also been implicated in the processing of complementary double strand ends in a recombination pathway of double strand break repair. More specifically, it was recently proposed that the 5Ј to 3Ј-exonuclease activity of EXO-1 partially compensates for a deficiency in the function of the Rad50⅐Mre11⅐Xrs2 complex in homologous recombination repair (32,34). Although a precise role of the WRN protein in a specific recombination pathway remains to be defined, there are several lines of evidence suggesting that a role of WRN in recombination is likely. 1) WS cells display severe genomic instability, elevated homologous recombination (9), and aberrant mitotic recombination (10,49). 2) the human WRN gene can suppress the increased homologous and illegitimate recombination in the yeast WRN homolog sgs1 (50). 3) WRN helicase can recognize and unwind a number of DNA structures with junctions (42) and catalyzes branch fork migration of Holliday junctions (51), a key DNA intermediate of homologous recombination. The results presented in this report suggest that WRN may function to modulate the cleavage activity of EXO-1 which is thought to be essential to the role of the enzyme in homologous recombination.
EXO-1 has been implicated in mismatch repair by its requirement in mutation avoidance (30) and interactions with mismatch repair proteins (33,35). Very recently, EXO-1 was demonstrated to be required in mismatch repair in a reconstituted system using extracts from human cells (45). Human EXO-1 was shown to perform mismatch-provoked excision directed by a strand break located either 5Ј or 3Ј to the mispair. Evidence that WRN protein may play a role in mismatch repair was provided by the demonstration that extracts from three WS fibroblastoid cell lines were deficient in repair of base-base and insertion/deletion mismatches (46). However, in the same study, extracts from four WS lymphoblastoid cell lines were all proficient in mismatch repair, suggesting that WRN protein could have a cell type-and/or tissue-specific role in mismatch repair. It was reported subsequently that introduction of human chromosome 8 encoding the WRN gene to WS fibroblast cell lines deficient in mismatch repair resulted in expression of WRN protein but did not restore mismatch repair activity (52). Thus, it is unclear whether or not WRN plays a direct role in mismatch repair. Based on our results that WRN and EXO-1 interact functionally, it is possible that WRN modulates EXO-1 activity in a pathway of mismatch repair. The elevated frequency of spontaneous mutations in the HPRT gene (4) and the pronounced genomic instability of WS cells (2, 3) may reflect at least in part a defect in a mutation avoidance pathway that relies on the mismatch repair machinery, including EXO-1.
It has been suggested that EXO-1 plays a role in Okazaki fragment processing in yeast that is at least partially redundant to FEN-1 (RAD27). On a biochemical level, EXO-1 is able to remove RNA primers from Okazaki fragment model sub-strates efficiently, suggesting that the enzyme may function in RNA primer removal during lagging strand synthesis (30). The ability of human EXO-1 to suppress the lethality of a rad27 mutant at the restrictive temperature is consistent with this notion (30). WRN stimulation of EXO-1 5Ј-flap cleavage may serve a function during Okazaki fragment maturation or the processing of other DNA structures at the replication fork. Evidence presented here demonstrates that WRN is able to stimulate EXO-1 cleavage of flap structures with 5Ј-ssDNA tails up to at least 10 nt, an activity that may be important in a FEN-1 independent pathway of 5Ј-flap processing (53). Alternatively, EXO-1 may play a role in the removal of the terminal ribonucleotide at the RNA-DNA junction during Okazaki fragment processing in a WRN-stimulated reaction. WRN may serve as a general factor to promote replication intermediate processing through its interaction with structure-specific endonucleases.
Very recently it was reported that EXO-1-dependent ssDNA at telomeres is required for cell cycle arrest in budding yeast ku70 deletion mutants (54). The observation that EXO-1 affects the metabolism of damaged telomeres and checkpoint responses is interesting in light of circumstantial evidence that WRN may also play a role in telomere metabolism. WS fibroblasts that express a transfected human telomerase (hTERT) gene have an extended lifespan and can become immortalized (55), suggesting that the absence of WRN protein confers accelerated cellular senescence in a pathway dependent on telomerase. However, WRN may also function in an alternate telomerase-independent mechanism, as suggested by studies of the yeast WRN homolog Sgs1 (56,57). Although no direct role of WRN protein in telomere metabolism has been established, it is conceivable that WRN stimulation of EXO-1 nuclease activity in subtelomeric regions is important for the replication and stability of telomeres and/or the activation of the appropriate DNA damage and signal checkpoint pathways to tolerate damaged telomeres.