Biochemical Characterization of the DNA Substrate Specificity of Werner Syndrome Helicase*

Werner syndrome is a hereditary premature aging disorder characterized by genome instability. The product of the gene defective in WS, WRN, is a helicase/exonuclease that presumably functions in DNA metabolism. To understand the DNA structures WRN acts upon in vivo, we examined its substrate preferences for unwinding. WRN unwound a 3′-single-stranded (ss)DNA-tailed duplex substrate with streptavidin bound to the end of the 3′-ssDNA tail, suggesting that WRN does not require a free DNA end to unwind the duplex; however, WRN was completely blocked by streptavidin bound to the 3′-ssDNA tail 6 nucleotides upstream of the single-stranded/double-stranded DNA junction. WRN efficiently unwound the forked duplex with streptavidin bound just upstream of the junction, suggesting that WRN recognizes elements of the fork structure to initiate unwinding. WRN unwound two important intermediates of replication/repair, a 5′-ssDNA flap substrate and a synthetic replication fork. WRN was able to translocate on the lagging strand of the synthetic replication fork to unwind duplex ahead of the fork. For the 5′-flap structure, WRN specifically displaced the 5′-flap oligonucleotide, suggesting a role of WRN in Okazaki fragment processing. The ability of WRN to target DNA replication/repair intermediates may be relevant to its role in genome stability maintenance.

Werner syndrome is a hereditary premature aging disorder characterized by genome instability. The product of the gene defective in WS, WRN, is a helicase/exonuclease that presumably functions in DNA metabolism. To understand the DNA structures WRN acts upon in vivo, we examined its substrate preferences for unwinding. WRN unwound a 3-single-stranded (ss)DNA-tailed duplex substrate with streptavidin bound to the end of the 3-ssDNA tail, suggesting that WRN does not require a free DNA end to unwind the duplex; however, WRN was completely blocked by streptavidin bound to the 3-ssDNA tail 6 nucleotides upstream of the singlestranded/double-stranded DNA junction. WRN efficiently unwound the forked duplex with streptavidin bound just upstream of the junction, suggesting that WRN recognizes elements of the fork structure to initiate unwinding. WRN unwound two important intermediates of replication/repair, a 5-ssDNA flap substrate and a synthetic replication fork. WRN was able to translocate on the lagging strand of the synthetic replication fork to unwind duplex ahead of the fork. For the 5-flap structure, WRN specifically displaced the 5-flap oligonucleotide, suggesting a role of WRN in Okazaki fragment processing. The ability of WRN to target DNA replication/repair intermediates may be relevant to its role in genome stability maintenance.
The gene defective in WS, designated WRN, encodes a nuclear (17) 1,432-amino acid protein with the seven conserved motifs found in the RecQ family of DNA helicases (18). WRN is a DNA-dependent ATPase and utilizes the energy from ATP hydrolysis to unwind double-stranded (ds)DNA (19,20). Using a conventional helicase directionality substrate with 2 radiolabeled oligonucleotides annealed to opposite ends of a long ssDNA molecule, it was determined that WRN helicase unwinds dsDNA with a 3Ј 3 5Ј-polarity with respect to the single strand that it is inferred to bind (21). WRN is also a 3Ј 3 5Ј-exonuclease (21)(22)(23), consistent with the presence of three conserved exonuclease motifs homologous to the exonuclease domain of Escherichia coli DNA polymerase I and RNase D in the protein sequence (24). The catalytic activities of WRN suggest that the protein plays an important role in a pathway of genome stability, but precisely what pathway of DNA metabolism is defective in WS is not well understood. The involvement of three human RecQ family members (WRN, BLM, and RECQL4) in inherited disorders (WS, Bloom syndrome, and Rothmund-Thomson syndrome, respectively) characterized by genomic instability, premature aging, and cancer suggests that the helicase function is likely to be important in the molecular pathology of disease (for review, see Ref. 25). A reasonable hypothesis is that specific DNA substrates may be acted upon by the helicase during replication, repair, or recombination to confer genomic stability.
A role of WRN in DNA replication has been suggested by its recovery from cells in a replication complex (26) as well as its interactions with other proteins of the replication machinery which include replication protein A (21,27), flap endonuclease 1 (FEN-1) (28), polymerase ␦ (29,30), topoisomerase I (26), and proliferating cell nuclear antigen (26). Among the replication defects observed in cells from WS patients, the delayed S phase progression (8) suggests that WRN may play a direct role in the metabolism of certain structures that potentially interfere with the progression of the replication fork. Sequence-specific DNA structures such as triplexes and tetraplexes potentially block DNA synthesis. WRN is able to resolve these structures (31)(32)(33), and most recently it was shown that WRN can alleviate the pausing of yeast polymerase ␦ at sites of tetraplexes by eliminating the secondary structure (34). WRN also actively unwinds the four-stranded X structure (33), a model for the Holliday junction recombination intermediate. Recent models propose that Holliday junctions can arise at sites of stalled DNA replication (35,36), and WRN may play a role in the subsequent processing of this structure during replication restart, perhaps in a recombinational repair pathway.
To understand better the substrate specificity of WRN helicase, we have examined the ability of WRN to unwind a panel of defined duplex DNA substrates using an in vitro strand displacement assay. Helicase activity assays were performed to identify the DNA duplex substrates most efficiently unwound by WRN. This system enabled us to assay carefully WRN interactions with the substrate in a functional context and to determine which features of the DNA substrate are important for WRN to initiate the unwinding reaction efficiently. It was reported previously that oligonucleotide duplex substrates with both 3Ј-and 5Ј-(ss)DNA tails are unwound preferentially compared with similar substrates with only 3Ј-tails (32,33). One aspect of this study entailed defining the minimum length requirements for the 5Ј-and 3Ј-ssDNA tails of forked duplex substrates required by WRN for efficient unwinding. In addition, we have used helicase substrates with specifically positioned steric blocks to probe the preference of WRN for forked duplex structures.
The replication defects and hypersensitivity to certain DNAdamaging agents of WS cells suggest a direct role of the protein in the processes of replication and/or repair. One important DNA structural intermediate in both of these processes is the 5Ј-ssDNA flap substrate. Flap structures may arise during Okazaki fragment processing in replication or strand displacement during DNA repair pathways such as base excision repair. Recently, we demonstrated that WRN interacts with human FEN-1 (28), a DNA repair/replication enzyme that processes 5Ј-ssDNA flap structures. In the current study, we have shown that WRN efficiently unwinds the 5Ј-flap DNA substrate. By its action at flap structures, WRN may facilitate strand displacement by an advancing DNA polymerase and processing of unannealed 5Ј-ssDNA flaps.
The catalytic unwinding activity of WRN may be important at specific structures of the replication fork. Although the synthetic replication fork lacks the ssDNA loading region of a typical helicase substrate, WRN effectively unwound the structure in a specific manner by directing its helicase activity in the direction of the fork, leaving the two duplex arms intact. This characteristic feature of DNA unwinding by WRN might be important at the site of a stalled replication fork or in some other aspect of DNA metabolism at the site of new DNA synthesis. The preference of WRN to unwind dsDNA substrates with junctions suggests that WRN may act upon these structures during the processes of DNA replication, recombination, or repair. The cellular defects and genomic instability of WS may arise from persistent DNA structures that fail to be acted upon by WRN helicase.

Protein, Oligonucleotides, and DNA Substrates Used in This Work-
Recombinant WRN protein used in this study was purified as previously described (28,37). PAGE-purified oligonucleotides, obtained from Midland Certified Reagent Company, are listed in Table I. DNA duplex substrates were prepared as described previously (38) and are shown in Tables II and III. Helicase Assays-Helicase assay reaction mixtures (20 l) contained 30 mM Hepes pH 7.6, 5% glycerol, 40 mM KCl, 0.1 mg/ml bovine serum albumin, 8 mM MgCl 2 , 2 mM ATP, 10 fmol of DNA duplex substrate (0.5 nM DNA substrate concentration), and the indicated amounts of WRN. For helicase reactions containing streptavidin, 15 nM streptavidin was preincubated with the DNA substrate with all reaction components except WRN for 10 min at 37°C. Helicase reactions were initiated by the addition of WRN and then incubated at 37°C for 15 min unless otherwise indicated. WRN concentrations used were 0.19, 0.38, 1.9, and 3.8 nM monomer. Reactions were quenched with 10 l of loading buffer (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% (19:1 acrylamide:bisacrylamide) polyacrylamide gels except where indicated in the figure legends. Ra-  diolabeled DNA species in polyacrylamide gels were visualized using a PhosphorImager and quantitated using the ImageQuant software (Molecular Dynamics).
The percent helicase substrate unwound was calculated by the formula: % unwinding ϭ 100 ϫ (P/(S ϩ P)), where P is the product and S is the substrate. The values of P and S have been corrected after subtracting background values in the no enzyme and heat-denatured substrate controls, respectively. Helicase data represent the mean of at least three independent experiments with standard deviations shown by error bars.

RESULTS
To determine the DNA substrate requirements for efficient unwinding of B form dsDNA by WRN helicase, we have tested a series of related DNA substrates (0.5 nM final DNA substrate concentration) incubated with various concentrations of WRN protein using a strand displacement assay. The products of the helicase reaction were analyzed by native PAGE to determine which substrates were most efficiently unwound. Under the reaction conditions used for these studies, the DNA oligonucleotides, both those unwound and those remaining in duplex state, were not appreciably degraded by nuclease activity of WRN as judged by quantitation of DNA products resolved on denaturing gels (data not shown). Thus, the percent of DNA substrate unwound (% displacement) is determined by the amount of intact oligonucleotide released divided by the total amount of DNA substrate in the reaction.
WRN Helicase Activity on Forked DNA Duplexes with Increasing 5Ј-Tail Length-Our initial studies were focused on determining the importance of the length of the noncomplementary 5Ј-ssDNA tail of a forked dsDNA substrate for the efficiency of unwinding by WRN (Table II, substrates 1-5). Because WRN has been reported to unwind dsDNA with a 3Ј-ssDNA overhang, the length of the 5Ј-ssDNA tail may also be an important factor in the WRN helicase reaction. WRN-catalyzed unwinding of a 19-bp DNA duplex with a 25-nucleotide 3Ј-overhang (substrate 1) is relatively inefficient, with appreciable strand displacement (20 -25%) during the 15-min time course only detected at the highest WRN protein concentrations tested (1.9, 3.8 nM WRN) (Fig. 1). Thus a 3.8-fold greater amount of WRN monomer (1.9 nM) compared with DNA substrate (0.5 nM) is required for detectable unwinding, suggesting that unwinding is stoichiometric. Using a DNA substrate with the same duplex region but flanked by a single noncomplementary nucleotide on the terminus of the 5Ј-tail (substrate 2), WRN unwound a significantly greater amount of substrate, 40 and 58% displacement at WRN concentrations of 1.9 and 3.8 nM, respectively. Using a DNA duplex substrate with a 5-nucleotide noncomplementary 5Ј-ssDNA tail (substrate 3), the unwinding reaction was even more significantly improved. In this case, lower concentrations of WRN (0.19 and 0.38 nM) were sufficient to unwind 10 and 28% of the dsDNA substrate, respectively. At these protein concentrations, WRN unwound ϳ5.5-fold more of substrate 3 compared with substrates 2 and 1, indicating that the enzyme is more efficient at unwinding dsDNA substrates with a 5-nucleotide 5Ј-ssDNA flanking region compared with substrates with only a 1-nucleotide 5Јnoncomplementary tail or no tail at all. At the two highest concentrations of WRN (1.9 and 3.8 nM), substrate 3 bearing the 5-nucleotide 5Ј-ssDNA tail was efficiently unwound by WRN, achieving 70 -80% displacement. Slightly higher levels of unwinding were attained at the two highest WRN concentrations with substrate 4, which has the 10-nucleotide 5Ј-tail. A DNA duplex with a long 5Ј-ssDNA tail of 26 nucleotides (substrate 5) was unwound significantly better (ϳ12-fold at a WRN concentration of 0.18 nM) than the DNA substrate with only the 3Ј-ssDNA tail (substrate 1), clearly indicating the preference of WRN for the forked substrate. Substrate 5 was unwound better than the fork duplex with the 5-nucleotide (substrate 3) or 10-nucleotide (substrate 4) 5Ј-ssDNA tails at 0.19 and 0.38 nM WRN protein concentrations. At higher WRN concentrations (1.9, 3.8 nM), substrates 3, 4, and 5 were unwound nearly equally well, suggesting that a forked duplex with 5-10 nucleotides of 5Ј-ssDNA tail is optimally unwound by WRN. Importantly, the helicase data on this family of substrates clearly show that a forked duplex is unwound substantially better than a 3Ј-tailed duplex.
WRN Helicase Activity on Forked DNA Duplexes with Increasing 3Ј-Tail Length-Because the forked duplex is the preferred substrate for WRN helicase, we next addressed the importance of the length of the noncomplementary 3Ј-ssDNA tail of a DNA duplex substrate that is flanked by a long 5Ј-ssDNA tail of 26 nucleotides (Table II, substrates 5-8). A DNA substrate possessing only a single noncomplementary 3Ј-nucleotide (substrate 6) was essentially not acted upon by the WRN helicase ( Fig. 2), similar to a previously published report (21) and our own data that WRN does not unwind a duplex DNA substrate flanked by only a 5Ј-ssDNA tail. However, if the DNA substrate possessed a 5-nucleotide 3Ј-ssDNA tail (substrate 7), significantly greater amounts of DNA unwinding were detected at WRN concentrations of 0.19 -3.8 nM. At the highest WRN protein concentrations tested, 1.9 and 3.8 nM, substrate 7 was unwound 44 and 54%, respectively, a 20-fold increase compared with substrate 6. Using a 10-nucleotide 3Ј-tailed substrate (substrate 8), a significantly greater amount of DNA substrate was unwound compared with substrate 7 at WRN protein concentrations of 0.19 -3.8 nM. The maximal difference between substrates 7 and 8 was ϳ2-fold at a WRN concentration of 0.38 nM. The 25-nucleotide 3Ј-ssDNA tailed substrate (substrate 5) was unwound very similarly compared with substrate 8 at all WRN levels. Altogether the helicase data indicate that a 3Ј-ssDNA tail of 5 nucleotides is the minimal length for efficient unwinding of a forked substrate by WRN and that a 10-nucleotide 3Ј-ssDNA tail can be judged to be optimal.
The results from WRN helicase assays using forked duplex substrates with a long 25-nucleotide 3Ј-ssDNA tail and increasing lengths of 5Ј-ssDNA tail indicated that 5-10 nucleotides is the minimal length of 5Ј-ssDNA tail necessary for optimal unwinding (Fig. 1). Likewise, the results from WRN helicase assays using forked duplex substrates with a long 26-nucleotide 5Ј-ssDNA tail and increasing lengths of 3Ј-ssDNA tail indicated that a forked duplex with a 5-nucleotide 3Ј-ssDNA tail is unwound by WRN, but a 10-nucleotide 3Ј-ssDNA tail is necessary for optimal unwinding (Fig. 2). Taken together, these results would suggest that WRN might efficiently unwind a forked duplex substrate with 3Ј-and 5Ј-ssDNA tail lengths of 10 nucleotides. We tested WRN on such a forked duplex (substrate 9) and found this to be the case (Fig. 2). Approximately 43% of the substrate was unwound at a WRN concentration of 0.38 nM (Fig. 2). At WRN concentrations of 1.9 and 3.8 nM, 68 and 85% of the substrate was unwound, respectively. WRNcatalyzed DNA unwinding was dependent on ATP hydrolysis because the poorly hydrolyzable analog ATP␥S failed to support the unwinding reaction (data not shown). These data indicate that WRN can efficiently unwind a forked duplex with 3Ј-and 5Ј-ssDNA tails of 10 nucleotides. Comparable levels of unwinding were detected for substrates 4, 5, 8, and 9, suggesting that a forked duplex with 3Ј-and 5Ј-ssDNA tails of 10 nucleotides is unwound as efficiently as the duplex with tails of 25 nucleotides (3Ј) and 26 nucleotides (5Ј).
DNA Structural Elements Important for WRN Helicase Activity-The reported 3Ј 3 5Ј directionality of WRN helicase (21) may be a consequence of unidirectional translocation on ssDNA in a 3Ј 3 5Ј direction; alternatively, the polarity of DNA unwinding may be determined by the binding specificity of WRN protein for a ssDNA/dsDNA junction characterized by a 3Ј-ssDNA overhang. The latter explanation was recently proposed to be responsible for the 3Ј 3 5Ј directionality of a recombinant Sgs1 protein fragment (39). To address this issue for WRN, we examined DNA binding by WRN to a number of DNA substrates used in this study, including the 3Ј-tailed duplex (substrate 1) and the forked duplex (substrate 5). Gel shift analysis of protein-DNA mixtures that had been incubated in the presence of ATP␥S or in the absence of nucleotide did not demonstrate a major level of stable binding to any of the duplex DNA substrates (data not shown). We were able to detect reproducibly 1-2% of the forked duplex substrate 5 molecules crosslinked to 3.8 nM WRN when the samples were treated with 0.25% glutaraldehyde after the binding incubation (data not shown). Under these conditions, WRN consistently failed to be cross-linked to substrates 1 or 2, suggesting that WRN may bind slightly preferentially to forked structures compared with the 3Ј-tailed substrate lacking a flanking 5Ј-tail (substrate 1) or bearing a single nucleotide 5Ј-tail (substrate 2), as detected by this assay.
To probe the importance of DNA structural elements for the WRN unwinding function, we tested WRN helicase activity on dsDNA substrates with specifically positioned biotin-streptavidin complexes (Table III, substrates 10 -12 and 19). The streptavidin-biotin complex is an extremely strong interaction (K d ϳ 10 Ϫ15 M) (40), and the diameter of a streptavidin tetramer, ϳ45 Å (41), has previously been shown to block dsDNA unwinding by T7 Gene 4 and DnaB helicases when it is positioned on the single strand on which the enzyme translocates (42,43).
Initially, we tested a substrate similar to the DNA duplex with a 25-nucleotide 3Ј-ssDNA overhang (substrate 1) which also has a biotin moiety conjugated to the terminal 3Ј-nucleotides of the ssDNA overhang (substrate 10). Streptavidin was effectively bound to the DNA substrate throughout the time course of the experiment as detected by a gel-shifted species when streptavidin was present (data not shown). WRN unwound substrate 10 nearly equally well in the presence or absence of streptavidin, as demonstrated by only a small reduction in the level of unwinding when streptavidin was present (Fig. 3). These results suggest that WRN is not required to load onto the end of the protruding 3Ј-ssDNA of the DNA substrate to unwind the duplex. We next tested the same 3Ј-tailed substrate, only the biotin was positioned 6 nucleotides upstream of the ssDNA/dsDNA junction (substrate 11). In this case, WRN was blocked from unwinding the substrate in the presence of streptavidin (Fig. 4). In contrast to these results, WRN unwound up to 40% of the DNA substrate in the absence of streptavidin. The fact that streptavidin only had a minor effect on WRN helicase activity on substrate 10 under the identical reaction conditions (Fig. 3) indicated that the strong inhibitory effect of streptavidin on the WRN helicase reaction is specific for substrate 11 and caused by the position of the bound biotin. We also tested WRN in a streptavidin displacement assay under reaction conditions identical to those of the helicase assay and found that WRN completely failed to displace streptavidin bound to the same biotin-conjugated oligonucleotide (T STEM 25BINT) used to construct substrate 11 (data not shown). These results suggest that streptavidin bound to the biotin-conjugate positioned 6 nucleotides upstream of the ssDNA/dsDNA junction interferes with the progression of WRN helicase into the duplex region. Because WRN was unable to displace streptavidin from ssDNA, the helicase inhibition by the internally positioned streptavidin complex suggests that streptavidin may have posed a block to translocation of WRN along the 3Ј-ssDNA tail to the ssDNA/dsDNA junction on   streptavidin bound in the presence or absence of WRN did not demonstrate any detectable difference when either ATP␥S was present or in the absence of nucleotide (data not shown). Likewise, we detected no difference in the DNase I footprint of substrate 1 in the presence or absence of WRN. Because we were unable to detect stable binding of WRN to DNA substrate 1 or 11 by either gel shift or DNase I footprint analyses (data not shown), we cannot exclude the possibility that the presence of streptavidin may have exerted a subtle effect on WRN binding to substrate 11, which contributed to the inability of the enzyme to unwind the streptavidin-bound substrate in the presence of ATP.
The DNA substrate with biotin positioned 6 nucleotides away from the junction provided 19 nucleotides of 3Ј-ssDNA tail upstream of the bound streptavidin for WRN to load. We were interested in the possibility that WRN might be able to bypass the steric block of streptavidin bound to the DNA substrate if the enzyme were presented with a fork structure. To address this, we tested WRN helicase activity on the forked duplex with 5Ј-and 3Ј-noncomplementary tails of 26 and 25 nucleotides, respectively, which also had the biotin moiety positioned on the 3Ј-ssDNA tail 6 nucleotides away from the junction (substrate 12). Gel shift analysis indicated that the streptavidin remained bound to the forked duplex substrate throughout the 60-min time course (data not shown). WRN unwound the biotin-conjugated fork substrate in the presence or absence of streptavidin (Fig. 4). The displaced strand retained the bound streptavidin (data not shown), indicating that WRN helicase did not dislodge the streptavidin from the forked substrate. The presence of 15-50 nM biotin during the 37°C 60-min incubation period of substrate 12 with WRN helicase did not result in dissociation of the streptavidin from the T STEM 25BINT oligonucleotide (data not shown), suggesting that streptavidin was not displaced by WRN during the unwinding reaction and able to rebind to the released T STEM 25BINT. A kinetic analysis of the DNA unwinding reaction demonstrated that unwinding of the streptavidin-bound forked substrate was reduced by 4-fold at the 1-min time point (Fig. 4). However, the percent of substrate 12 unwound in the presence or absence of streptavidin approached similar levels by 5 min. In contrast, WRN failed to unwind substrate 11 bound to streptavidin appreciably throughout the 60-min incubation. These results suggest that if WRN is presented with a DNA fork, the enzyme is able to bypass the steric block imposed by streptavidin bound to a similar DNA duplex substrate with only a protruding 3Ј-ssDNA tail. It is evident that WRN interacts differently with the forked duplex compared with the 3Ј-ssDNA tailed duplex to initiate unwinding. This difference is likely to be responsible for the preferential unwinding of the fork structure by WRN.
WRN Unwinds a 5Ј-Flap Substrate Flanked by Adjacent Duplex DNA-Having demonstrated that WRN preferentially unwinds forked DNA duplexes with noncomplementary 3Ј-and 5Ј-ssDNA tails compared with a DNA substrate that possessed only a 3Ј-ssDNA tail, we were interested in the possibility that WRN might unwind dsDNA substrates that possess a junction but lack either one or both of the preexisting ssDNA tracts flanking the duplex region. A DNA substrate that belongs to this class of molecules is the 5Ј-ssDNA flap substrate, i.e. a dsDNA substrate that has a 5Ј-ssDNA tail flanked immediately by an upstream DNA duplex as opposed to a 3Ј-ssDNA tail of a traditional forked duplex substrate (Table III, substrates 14 -17). 5Ј-Flap substrates are particularly relevant in a biological context because they are proposed intermediates in the process of Okazaki fragment processing during lagging strand DNA replication and also in DNA repair pathways such as base excision repair.
To assess whether WRN is able to unwind a 5Ј-flap substrate, we tested a DNA substrate related to the forked duplex with 26-and 25-nucleotide 5Ј-and 3Ј-ssDNA tails, respectively (substrate 5), which also contains an upstream 25-mer hybridized to the upstream ssDNA region that resides below the flap (substrate 17). Substrate 17 is characterized by a nick separating the adjacent oligonucleotides. This substrate is cleaved efficiently by FEN-1 (44), 2 whereas a substrate with a single nucleotide gap between the two adjacent duplexes is a poor substrate for FEN-1 (44). The results from a typical WRN helicase assay using substrate 17 with a radioactive label on the 5Ј-end of the 5Ј-flap oligonucleotide is shown in Fig. 5A. As shown in lane 2, 3.8 nM WRN efficiently unwound the 5Ј-flap substrate, releasing the labeled 5Ј-flap oligonucleotide (FLAP26) as evidenced by its comigration with the heat-denatured substrate control (lane 3). WRN-catalyzed unwinding of the flap substrate is dependent on ATP hydrolysis because the poorly hydrolyzable analog ATP␥S failed to support helicase activity (lane 4).
The observation that WRN does not require a preexisting free 3Ј-ssDNA tail adjacent to the duplex region for WRN to unwind the substrate raised the possibility that WRN helicase may act upon the upstream annealed primer to unwind the oligonucleotide thereby generating a forked duplex that would be unwound efficiently by WRN. This is not the case because WRN did not release the labeled upstream oligonucleotide (U25) annealed to the substrate (Fig. 5B, lane 2). Rather, WRN displaced the 5Ј-flap oligonucleotide, leaving the duplex species still containing the upstream oligonucleotide (Fig. 5B, lane 2). Analyses of the labeled upstream oligonucleotides on denaturing polyacrylamide gels revealed that the upstream primer was Ͼ95% intact (data not shown), indicating that it was not degraded to any significant amount by the 3Ј 3 5Ј-exonuclease activity of WRN. These results indicate that the helicase activity of WRN protein specifically releases the 5Ј-flap oligonucleotide, leaving the upstream duplex intact.
To characterize further the WRN helicase reaction on a 5Јflap DNA duplex substrate, we investigated the dependence of WRN helicase activity on the length of the 5Ј-flap oligonucleotide in the DNA substrate (Table III, substrates [13][14][15][16][17]. The results demonstrate that a 1-nucleotide 5Ј-flap (substrate 14) was unwound only slightly better than a nicked DNA duplex (substrate 13) at the highest WRN protein concentrations tested (1.8 and 3.6 nM) (Fig. 6). WRN was able to unwind a 5-nucleotide 5Ј-flap substrate (substrate 15) significantly better, unwinding 25 and 40% of the substrate at WRN concentrations of 1.8 and 3.6 nM, respectively. The forked duplex containing a 5-nucleotide 5Ј-ssDNA tail (substrate 3) was unwound more efficiently than the related flap substrate (substrate 15) at WRN concentrations of 0.18 -3.6 nM, with the greatest difference, 4-fold, at 0.36 nM. A 5Ј-flap substrate with 10 nucleotides in the 5Ј-ssDNA region (substrate 16) was unwound better than the 5-nucleotide flap substrate (substrate 15), particularly at the highest WRN concentration, 3.6 nM. At 3.6 nM WRN, substrates 16 and 17 (the 26-nucleotide 5Ј-flap structure) were unwound to a similar extent. A comparison of the unwinding data from experiments using substrate 17, 5 (the related forked duplex), and 1 (the 3Ј-tailed duplex) indicates that substrates 5 and 17 were unwound significantly more efficiently by WRN than substrate 1. The forked duplex was unwound somewhat more efficiently by WRN than the flap substrate (compare substrates 5 versus 17), with nearly 2-fold more of the substrate unwound at a WRN concentration of 0.36 nM. These results indicate that the presence of the upstream primer is slightly inhibitory to the WRN helicase reaction; nonetheless, the flap substrate is unwound relatively well by WRN. Altogether, these data indicate that WRN is able to unwind duplex substrates with a 5Ј-ssDNA flap of 5 nucleotides. DNA substrates with longer 5Ј-ssDNA flaps (10, 26 nucleotides) are unwound significantly better by WRN than the 5-nucleotide 5Ј-flap. Importantly, a 5Ј-ssDNA flap substrate was preferred by WRN over a DNA duplex with just a 3Ј-ssDNA overhang, indicating that the helicase does not require a free 3Ј-ssDNA tail adjacent to the duplex region to catalyze efficient unwinding of dsDNA.
Biotin-Streptavidin Complex Fails to Block WRN Unwinding of the 5Ј-Flap Substrate-The ability of WRN to bypass the steric block to DNA unwinding imposed by streptavidin bound to 3Ј-ssDNA tail of a forked duplex substrate suggested that WRN might behave similarly on the 5Ј-flap structure. We addressed this question by examining WRN helicase activity on flap substrates with streptavidin bound to the terminal 3Јnucleotide (substrate 18) or 6 nucleotides upstream from the junction (substrate 19), as described previously for the forked duplex substrates. The results of these studies are presented in Fig. 7. The presence of streptavidin bound to either position partially inhibited the ability of WRN to displace the 5Ј-flap oligonucleotide; however, WRN helicase activity was evident on the streptavidin-bound flap substrates, even at the earliest time point of 1 min (Fig. 7). WRN was able to effectively release the 5Ј-ssDNA flap oligonucleotide, leaving streptavidin bound to the remaining duplex, as evidenced by its comigration with the control streptavidin-bound duplex (U25⅐T STEM 25B3 or U25⅐T STEM 25BINT) (data not shown). Quantitative analyses of the unwinding data indicated that WRN helicase activity reached a plateau at 10 min for both substrates in the presence or absence of streptavidin (Fig. 7). At the 10-min time point, WRN unwound ϳ90 and 75% of substrate 18 in the absence and presence of streptavidin, respectively. For substrate 19, WRN unwound 87 and 61% of the flap substrate in the absence and presence of streptavidin, respectively, at the 10-min time point. These results suggest that if WRN is presented with a DNA flap substrate, the enzyme is able to bypass the steric block imposed by streptavidin bound to a similar DNA duplex substrate with only a protruding 3Ј-ssDNA tail (substrate 11). We also tested WRN helicase activity on 5Ј-flap substrates in which streptavidin was bound to biotin positioned at the 5Јterminal nucleotides of the 5Ј-flap oligonucleotide (T STEM 25B) or the 3Ј-terminal nucleotides of T STEM 25 (T STEM 25B3). For both of these substrates, WRN effectively displaced the 5Ј-flap oligonucleotide in the absence or presence of streptavidin (data not shown), suggesting that WRN does not utilize entry points from the 5Ј-end of the flap oligonucleotide or the downstream blunt duplex end to unwind the FLAP26⅐T STEM 25 duplex. Importantly, the ability of WRN to bypass the internal streptavidin block on the 5Ј-flap substrate suggests that WRN interacts differently with the 5Ј-flap structure compared with the 3Ј-ssDNA tailed duplex to initiate unwinding. This difference, also observed for the forked duplex with 3Ј-and 5Ј-ssDNA tails, suggests that WRN recognizes the flap/fork structure by a different mechanism than the 3Ј-tailed duplex, resulting in greater unwinding.
WRN Unwinds a Synthetic Replication Fork DNA Substrate-The fact that 5Ј-flap DNA substrates were unwound relatively efficiently by WRN suggested that WRN protein contacts with preexisting ssDNA, at least in the upstream 3Ј-DNA flanking region of the substrate, are not required for WRN helicase activity on duplex DNA. To address whether WRN protein contacts with ssDNA in the 5Ј-flanking region adjacent to the DNA duplex of a fork substrate are important, we tested a forked substrate with a 3Ј-ssDNA tail and a 5Ј-dsDNA tail (substrate 20). As shown quantitatively in Fig. 8, substrate 20 was unwound, indicating that WRN does not require that the 5Ј-element of the forked duplex be single-stranded to unwind the substrate. Substrate 20 was unwound as efficiently as the 5Ј-ssDNA flap substrate (substrate 17), and the two substrates displayed a reduced level of unwinding compared with the forked duplex with 5Ј-and 3Ј-ssDNA tails (substrate 5), with the greatest difference observed at 0.19 nM WRN (Fig. 8).
We also tested a forked duplex in which both the 5Ј-and 3Ј-tails were double-stranded (substrate 21). This substrate mimics at the DNA level a synthetic replication fork with double-stranded leading and lagging strands. The fully duplex forked DNA substrate (substrate 21) was unwound by WRN at the two highest concentrations tested (1.9 and 3.8 nM) (Fig. 8).
Interestingly, the combined reduction in unwinding of substrates 17 and 20 compared with substrate 5 approximately represented the observed reduction in unwinding of substrate 21 (Fig. 8). Nonetheless, WRN was clearly able to unwind the synthetic replication fork because nearly 70% of substrate 21 was unwound at a WRN concentration of 3.8 nM. To examine the possibility that the double-stranded regions of the 5-tail (substrates 20 and 21) or 3Ј-tail (substrate 21) might also be unwound by WRN, we tested these substrates in which the radioactive label was on the oligonucleotides FLAP26COMP or U25 (Table I), respectively. In either case, the flanking duplex adjacent to the forked duplex was not unwound (data not shown), indicating that the WRN helicase reaction was specifically directed toward unwinding the synthetic replication fork in the direction of the fork.

DISCUSSION
In this study, we have examined the DNA substrate requirements for efficient unwinding of duplex DNA substrates by WRN helicase. Previous studies had suggested that WRN unwound forked DNA duplexes with both 3Ј-and 5Ј-noncomplementary ssDNA tails better than a DNA duplex that has only a protruding 3Ј-ssDNA tail (32,33). However, the minimal substrate requirements for efficient unwinding by WRN had not been investigated previously. It was reported that WRN binds preferentially to ssDNA over dsDNA by ϳ5-fold (37), but the source of DNA for that study was single-stranded and double-stranded pKS plasmid DNA and provides only limited information on DNA binding. DNA requirements for a functional biochemical activity of WRN were investigated here.
Our results indicate that WRN unwinds forked duplex substrates up to 12-fold greater than a simple 3Ј-ssDNA tailed duplex substrate, suggesting that DNA structural elements may play a role in the preferential unwinding of forked duplexes by WRN. A length of 10 nucleotides in either the 3Ј-or 5Ј-ssDNA tail provides an optimal forked duplex substrate for unwinding by WRN; however, DNA duplex substrates with shorter 3Ј-and 5Ј-ssDNA tails can be unwound by WRN. This information can be compared with that of bacteriophage T7 Gene 4, a well characterized hexameric 5Ј 3 3Ј-helicase that unwinds a forked DNA substrate by a mechanism in which the 5Ј-tail of the forked duplex substrate passes through the central channel of the helicase (45). For T7 DNA helicase, a minimal 3Ј-ssDNA tail of 10 nucleotides was required for efficient unwinding (46), a parameter comparable with that of WRN. A 10-nucleotide 5Ј-ssDNA tail of a fork duplex was not unwound by T7 helicase; however, unwinding rates increased gradually with 5Ј-tails greater than 13 nucleotides up to 65 nucleotides, the longest tail tested (45). The minimal loading site on the 3Ј-ssDNA tail of the fork required by WRN to unwind the fork duplex, 5 nucleotides, may be less than that determined for the 5Ј-ssDNA loading site of T7 helicase; however, differences in DNA substrates and reaction conditions make a direct comparison difficult. BLM also prefers a fork compared with a 3Ј-tailed duplex for unwinding (33), although the minimal lengths of the ssDNA tails required for efficient unwinding have not been determined. It will be of interest to determine whether other human RecQ family helicases exhibit similar or different substrate preferences compared with WRN.
Although a number of helicases prefer a forked duplex compared with a duplex with a ssDNA tail for unwinding (e.g. Refs. 43 and 46), this is not a universal property of all DNA helicases. For example, an Sgs1 recombinant protein fragment, containing the conserved RecQ helicase domain but lacking the aminoand carboxyl-regions of the full-length protein, preferentially unwound a DNA duplex with an 18-nucleotide 3Ј-ssDNA tail compared with a forked duplex with a protruding 3Ј-ssDNA tail as well as a noncomplementary 19-nucleotide 5Ј-ssDNA tail (39). The apparent difference in helicase substrate preference (forked duplex versus 3Ј-ssDNA tailed duplex) between the recombinant Sgs1 fragment and WRN may reflect an intrinsic property of the enzymes. Alternatively, the recombinant Sgs1 fragment (amino acid residues 400 -1268), which lacked 399 amino-terminal residues and 179 carboxyl-terminal residues, may exhibit properties distinct from that of full-length Sgs1 protein, including its substrate specificity for unwinding. Further studies are necessary to compare the helicase properties and substrate preferences of native RecQ helicases.
The directionality of movement of WRN helicase along ssDNA has been inferred from strand displacement assays using a linear partial duplex DNA substrate containing 2 radiolabeled oligonucleotides that are annealed to the very proximal opposite ends of a long ssDNA molecule (21). A long ssDNA region of several thousand nucleotides separates the annealed oligonucleotides. Preferential release of 1 of the 2 oligonucleotides on such a substrate is used to define the polarity of movement of the helicase along the bound ssDNA residing between the duplexes. Using this approach, WRN was shown to be a 3Ј 3 5Ј-helicase (21), a property shared by all RecQ helicases characterized to date (47). Consistent with this directionality, it was shown that WRN unwinds a duplex DNA substrate with a 3Ј-ssDNA tail but not a substrate with a 5Ј-ssDNA tail (Ref. 21 and this paper). These findings might lead one to believe that WRN requires a free 3Ј-ssDNA tail to initiate unwinding of at least duplex DNA substrates. The ability of WRN to unwind 5Ј-flap and synthetic replication fork substrates indicates that WRN does not require a preexisting 3Ј-ssDNA tail in the helicase substrate for either loading or initiation of unwinding.
To characterize the DNA structural elements recognized by WRN to initiate unwinding of dsDNA, we tested the ability of the helicase to unwind DNA substrates with a specifically positioned steric block in the 3Ј-ssDNA strand flanking a duplex. WRN does not require a free 3Ј-end to load onto the 3Ј-tailed substrate, but it is completely blocked by a streptavidin complex bound to biotin conjugated to the 3Ј-tail 6 nucleotides upstream of the ssDNA/dsDNA junction. This impediment may be the result of the inability of WRN to translocate 3Ј 3 5Ј along the ssDNA across the block to the junction. Alternatively, WRN may fail to recognize the ssDNA/dsDNA junction because the streptavidin-bound species masks the junction. When WRN is presented with the forked version of this substrate with streptavidin bound to the biotin 6 nucleotides upstream, the helicase effectively unwinds the substrate. This indicates that WRN is able to bypass the steric block in the context of the fork junction. A similar observation was made for the 5Ј-ssDNA flap substrate. We would suggest that WRN preferentially recognizes the fork-flap structure to commence unwinding over simply tracking along the free 3Ј-ssDNA tail. Thus, WRN may utilize two distinct modes to initiate DNA unwinding, depending on the type of substrate it confronts. The preferential unwinding of fork-flap duplex substrates compared with the duplex substrate with only a 3Ј-ssDNA tail suggests that the ability of WRN to recognize the junction of the former structures selectively increases the enzyme's efficiency to initiate DNA unwinding.
Our helicase studies with a synthetic replication fork substrate demonstrate that unwinding specifically occurs in the direction of the fork and that neither the double-stranded 3Ј-or 5Ј-arms of the fork are unwound. If WRN helicase retains the same 3Ј 3 5Ј directionality of translocation during the unwinding reaction of the synthetic replication fork as it is proposed to have on ssDNA based on results from experiments with a helicase directionality substrate (21), it may be inferred that WRN tracks along the lagging strand template in a 3Ј 35Ј direction. The unidirectional movement of WRN in a 3Ј 3 5Ј direction and disruption of the complementary Watson-Crick bp of the duplex region ahead of the fork structure would result in the release of the complementary DNA species, whether it is a single strand or a partially duplex molecule that retains a double-stranded region within the arm of the original DNA substrate.
The ability of WRN to unwind 5Ј-flap and synthetic replication fork substrates efficiently suggests that WRN has the capacity to load onto duplex DNA substrates by virtue of a junction recognition property of the enzyme. Several other well characterized DNA helicases, such as E. coli DnaB (43) and bacteriophage T7 Gene 4 (46), strongly prefer forked DNA substrates compared with duplex substrates with only a 5Ј-or 3Ј-ssDNA tail for unwinding. However, chemical modifications of the noncomplementary tails may differentially affect the rate of unwinding, depending on the helicase. In the case of DnaB, a 5Ј 3 3Ј-hexameric helicase, introduction of modifications (abasic sites, hexethylene glycol 1-phosphate groups) to the 5Ј-tail of a forked duplex substrate negatively affects unwinding, whereas no effect was observed if the modification was introduced to the 3Ј-tail (43). Interestingly, if the 5Ј-arm of the forked duplex is rendered double-stranded, the substrate is much less efficiently unwound by DnaB, suggesting that the 5Ј-ssDNA tail serves as an initial binding site for DnaB. In contrast, DnaB rapidly unwinds forked duplex DNA if the 3Ј-tail is double-stranded (43). The results from these studies are consistent with a proposed model for DnaB helicase function in which the 3Ј-tail of a forked duplex stimulates unwinding by sterically determining whether one of the two strands is excluded from the central channel of the DnaB, a requirement for the unwinding reaction. T7 Gene 4 helicase, a 5Ј 3 3Јhelicase, also requires a forked duplex for efficient unwinding. Gene 4 helicase requires ssDNA regions in both the 3Ј-and 5Ј-tails to initiate unwinding (46), suggesting that helicase protein contacts with both of the ssDNA tails in the forked duplex substrate are important to initiate unwinding. This is not the case for WRN because the synthetic replication fork substrate (substrate 21) was detectably unwound by WRN. Thus, WRN may initiate unwinding of DNA duplex substrates by a mechanism that does not require an interaction between the protein and the ssDNA tails of the forked duplex substrate.
The observation that a 3Ј-ssDNA tail is not a DNA structural requirement for unwinding of B form duplex DNA by WRN was reported for certain other DNA structures. WRN is active as a helicase on a synthetic X structure (a model for the Holliday junction recombination intermediate) as well as a blunt ended duplex containing a centrally located 12-nucleotide singlestranded "bubble" (33). The latter structure contains ssDNA in the bubble region, whereas the former structure forms a stacked X structure in the presence of Mg 2ϩ (present during the helicase assay) and therefore lacks ssDNA, even at its core. Thus WRN certainly does not require a free 3Ј-ssDNA tail or any preexisting ssDNA tracts in the substrate to catalyze efficient unwinding, a result consistent with the findings reported here.
The action of WRN helicase on the 5Ј-flap substrate is interesting from a biological standpoint because this structure is a key DNA intermediate during lagging strand synthesis of DNA replication. The average size of an Okazaki fragment (100 -150 nucleotides) (48) may be influenced by the action of WRN helicase activity on the flap structure. In this study, we have demonstrated that WRN is able to displace the 5Ј-ssDNA flap, even when the preexisting 5Ј-ssDNA flap of the helicase sub-strate is a relatively short length of 5-10 nucleotides. WRN helicase may regulate the size of the 5Ј-ssDNA flap of the Okazaki fragment during strand displacement synthesis by a DNA polymerase. The physical (30) and functional (29) interaction between WRN protein and polymerase ␦ may be well suited to coordinate displacement synthesis by polymerase ␦. The robust ability of WRN to stimulate the FEN-1 cleavage reaction on 5Ј-flap structures (28) may also play an important role in Okazaki fragment processing because FEN-1 is implicated in the process of DNA replication (49,50). Formation of the 5Ј-ssDNA flap by the combined action of strand displacement by a DNA polymerase and helicase is likely to be well coordinated with processing of the flap structure by an endonuclease such as FEN-1. The physical and functional interaction between WRN and FEN-1 suggests a provocative role of WRN during replication. The replication defects observed in WS cells (3,(5)(6)(7)(8) may reflect abnormal processing of specific structures associated with the replication fork. A current model for Okazaki fragment processing in yeast suggests that the displaced 5Ј-ssDNA flap species is subsequently acted upon by the concerted action of two endonucleases, Dna2 and FEN-1 (48), resulting in the excision of the flap so that ligation may take place. WRN may participate in the maturation of Okazaki fragments in human cells by virtue of its helicase activity on flap structures and functional interactions with DNA polymerase ␦, replication protein A, FEN-1, and perhaps other proteins associated with the replication fork. The increased spontaneous mutagenesis, genomic instability, and impaired S phase progression in WS cells and the identification of WRN in a replication complex (26) support the hypothesis that a defect in replication contributes to the WS cellular phenotypes.
WRN may have a specialized role in assisting in the resumption of replication after replication fork pausing/arrest. In support of this notion, WRN translocates to sites of replication/ repair when replication is perturbed by hydroxyurea (51,52). The ability of WRN helicase to unwind the duplex ahead of a fork structure that contains fully duplexed leading and lagging strands may be important for the maintenance of fork progression when replication undergoes pausing as in the case of nucleotide depletion or arrest at the sites of DNA damage. Precisely how WRN acts in vivo to maintain genome stability remains to be defined, but it is highly likely to involve its helicase function on a structural intermediate of an important DNA metabolic pathway.