DNA secondary structure of the released strand stimulates WRN helicase action on forked duplexes without coordinate action of WRN exonuclease

Abstract

Werner syndrome (WS) is an autosomal recessive premature aging disorder characterized by aging-related phenotypes and genomic instability. WS is caused by mutations in a gene encoding a nuclear protein, Werner syndrome protein (WRN), a member of the RecQ helicase family, that interestingly possesses both helicase and exonuclease activities. Previous studies have shown that the two activities act in concert on a single substrate. We investigated the effect of a DNA secondary structure on the two WRN activities and found that a DNA secondary structure of the displaced strand during unwinding stimulates WRN helicase without coordinate action of WRN exonuclease. These results imply that WRN helicase and exonuclease activities can act independently, and we propose that the uncoordinated action may be relevant to the in vivo activity of WRN.

1. Introduction

Werner syndrome (WS) is a human autosomal recessive disorder caused by mutations in the WRN gene [1] and characterized by premature aging and increased incidence of cancer [2]. WS cells show genome instability, such as various types of chromosomal aberrations [3], accelerated replicative senescence [4], and accelerated telomere loss [5]. In addition, WS cells are sensitive to many different types of DNA damaging agents, such as 4-nitroquinoline 1-oxide (4-NQO), camptothecin (CPT), and inter-strand crosslinks (ICLs) [6–8].

WRN is a member of the RecQ helicase family and has both heli-case and exonuclease activities [9,10]. WRN helicase unwinds a variety of DNA structures [11,12]. WRN exonuclease degrades nucleotides from 3′-recessed termini, gaps, nicks, and blunt ends of forked duplexes [12,13]. The two activities of WRN are modulated by interacting proteins that are involved in DNA repair, replication, and/or recombination pathways [14–17], thus suggesting that WRN plays an important role in these DNA transactions.

WRN helicase and exonuclease can simultaneously act on a single forked DNA duplex [13]. The WRN helicase unwinds at the fork of the substrate, while the WRN exonuclease digests at the blunt end. This coordinated action results in removing a DNA strand from the long, forked duplex that is not completely unwound by WRN helicase.

However, WRN can completely unwind 16- and 22-bp forked duplexes without digestion of the duplexes. A rapid unwinding of the short forked duplexes coincides with a lack of exonuclease progression with time. In addition, in the presence of RPA, WRN can unwind a 34-bp forked duplex and release the full-length strand without shortening the forked duplex [13]. RPA stimulation of WRN helicase thereby inhibits WRN exonuclease on the forked duplex and WRN helicase may act independently of WRN exonuclease.

Although the exact mechanism for the regulation of WRN exonuclease is not understood, one question raised by these studies is whether the stability of the remaining duplex during unwinding and the displaced strand can regulate WRN helicase and exonuclease activities.

In this study, we aimed to identify DNA structural elements that influence WRN activities. Our results herein reveal that a DNA secondary conformation, such as a hairpin loop of the displaced strand, leads WRN to unwind forked duplexes proficiently and that the location of the secondary structure with respect to the fork affects WRN helicase efficiency. Thus, this study suggests that WRN helicase can unwind forked duplexes without coordinating with the WRN exonuclease.

2. Materials and methods

2.1. Proteins and oligonucleotide substrates

Oligonucleotides were ordered from Integrated DNA Technologies (Coralville, IA, USA). For biochemical assays (see below), the indicated oligonucleotides in figures were radiolabeled at the 5′ end (5′ end-labeled) with [γ-³²P] ATP (3000 Ci/mmol, Amersham Radiochemicals) using T4 polynucleotide kinase (New England Bio-labs) for 30 min at 37 °C and heat-inactivated for 10 min at 95 °C. Forked duplex substrates were formed by annealing the labeled oligonucleotides to their unlabeled complementary strands at a molar ratio of 1:2 by incubation, as previously described [13].

Recombinant wild-type WRN and X-WRN containing a point mutation (E84A) in the exonuclease domain were expressed using a baculovirus/insect cell expression system and purified, as described previously [18]. The recombinant glutathione S-transferase (GST)-tagged WRN fragment, HlcRQC, was purified, as described previously [19]. Protein concentrations were determined by a Bio-RAD assay (BIO-RAD) using BSA as the standard.

2.2. WRN helicase and exonuclease reactions

DNA substrates were mixed in reaction buffer (10–20 μl, 40 mM Tris–Cl, pH 8.0, 4 mM MgCl₂, 5 mM DTT, 2 mM ATP, and 0.1 mg/ml BSA) or ‘exo-free’ buffer (10–20 μl, 50 mM Hepes–KOH, pH 7.5, 2 mM MgCl₂, 5 mM DTT, 2 mM ATP, and 0.1 mg/ml BSA). In all cases, the concentration of DNA substrates was 1.0 nM. Reactions were initiated by the addition of proteins to the reaction mixture and incubated for 15 min at 37 °C.

2.3. Analysis of helicase and exonuclease products

Reactions were terminated by the addition of 3× stop dye (50 mM EDTA, 40% glycerol, 0.9% SDS, 0.1% bromophenol blue, and 0.1% xylene cyanol) to a 1× final concentration, along with a 100-fold molar excess of an unlabeled competitor oligonucleotide (identical to the labeled oligonucleotide strand of forked duplexes) to prevent reannealing of the unwound-radiolabeled ssDNA products. Helicase products were analyzed on native polyacrylamide gels, and visualized using a PhosphorImager (Typhoon 9400, GE Life Science).

Reactions were terminated by an addition of equal volume of formamide stop dye (80% formamide, 0.5× Tris borate, 0.1% bromophenol blue, and 0.1% xylene cyanol). Reaction products were heat-denatured for 5 min at 95 °C, analyzed on 14% sequencing polyacrylamide gels, and visualized using a PhosphorImager.

2.4. DNA secondary structure prediction

The potential secondary structure for all DNA oligonucleotides was found using the structure prediction program, RNAfold, at rna.tbi.unive.ac.at/cgi-bin/RNAfold.cgi (web-based program). Structure predictions were determined using DNA parameters and the default options in the program.

2.5. Native gel electrophoresis

Single-stranded DNA oligonucleotides were 5′-radiolabeled, as described in the oligonucleotide substrates above. DNA oligonucleotides were then incubated for 15 min at 37 °C in buffer plus 10% glycerol. DNA oligonucleotides (1 pmol) were analyzed on a 10% nondenaturing polyacrylamide gel (19:1 acrylamide/bis) and electrophoresed at 160 v in 1× TBE for 2.5 h at room temperature. Radiolabeled DNA species were visualized using a PhosphorImager.

3. Results

3.1. Sequence of the DNA substrates affects WRN enzymatic activities

A WRN digestion analysis of a forked duplex containing 4 telomeric repeats (TTAGGG) noticeably showed short products of 39-, 32-, and 26 nt-long fragments (Fig. 1A) as shown in a previous report [13]. The corresponding DNA bases to these fragments were either at the end of or within a homopolymeric run of three 3Gs (GGG in the 34-bp forked duplex, labeled as 3G), as shown in Fig. S1. This observation suggests that WRN exonuclease activity may be affected by DNA sequence context.

Fig. 1.

DNA sequences on forked duplexes influence both WRN exonuclease and helicase activities. WRN protein was incubated with the 34-bp forked duplexes (1.0 nM) under the standard reaction conditions at 37 °C for 15 min. Reactions were terminated in the appropriate stop dye. (A) Analysis of wild-type WRN exonuclease activity. DNA substrates and 0 (lane 1), 1.7, 3.4, 6.7 ng WRN (lanes 2–4). Reaction products were run on a 14% denaturing polyacrylamide gel. Values indicate the product length and extent of degradation. (B) Analysis of wild-type WRN helicase activity. DNA substrates and 0 (lane 1), 1.7 and 3.4 ng WRN (lanes 2 and 3). Reaction products were run on a 12% native polyacrylamide gel. Δ, heat-denatured substrate control (lane 4).

To examine the influence of the DNA sequence context on WRN exonuclease and helicase activities, we replaced GGG in the 34-bp forked duplex with GGGG (labeled as 4G) or CGCG (2CG), as shown in Fig. S1, and the three DNA substrates were incubated with increasing amounts of WRN. Reaction products were analyzed on native gels to show WRN helicase activity and on denaturing gels to visualize WRN exonuclease activity.

An analysis of the reaction products of the 4G substrate on a denaturing gel revealed that WRN degraded the 5′-labeled strands of the 4G substrate, starting from the blunt end (Fig. 1A). The pattern of digestion of the 4G substrate showed prominent bands of 38 and 31 nt in length (Fig. S1). In the 2CG substrate, prominent bands longer than 37 nt were predominantly detected, and the extent of digestion of the full-length 49-mer strand was less than that of the 3G and 4G substrates. These data indicate that the extent and pattern of WRN exonuclease digestion are affected by DNA sequence context.

Native gels showed that the distinctly displaced strand was shown in the 2CG substrate and migrated just below the labeled 49-mer ssDNA marker (2CG substrate; lanes 1–4 in Fig. 1B), indicating that WRN was able to unwind almost the entire full-length 49-mer strand. In contrast, the displaced strands of the 3G and 4G DNA substrates migrated much below the labeled 49-mer strand (3G and 4G substrates, lanes 1–4 in Fig. 1B) and appeared as smeared bands that indicate partially degraded single-stranded fragments. These data suggest that DNA sequence context affects WRN helicase unwinding and WRN exonuclease digestion.

3.2. WRN helicase acted independently of WRN exonuclease on a single substrate

Since WRN was able to unwind the full-length 49-mer of the 2CG forked duplex with very little digestion, we tested the influence of WRN exonuclease activity on the WRN-catalyzed unwinding of this forked duplex using an exonuclease-deficient WRN mutant and a WRN helicase fragment.

A WRN variant (X-WRN) that contains a single amino acid substitution (E84A) in the exonuclease domain of WRN results in a lack of exonuclease activity while preserving helicase activity [9]. The 2CG forked duplex was incubated with increasing amounts of X-WRN, and the reaction products were analyzed on a native gel to display the helicase activity. The gel showed that the full-length 49-mer of the 2CG forked duplex was displaced (Fig. 2). A recombinant WRN fragment (HlcRQC, amino acid 500–1104) containing the WRN helicase domain and RQC motif was incubated with the 2CG forked duplex. The full-length 2CG forked duplex was unwound, but the extent of unwinding was less than that by X-WRN. In contrast, the 3G forked duplex was not unwound by either X-WRN or HlcRQC (Fig. 2). These results suggest that DNA sequence context influences WRN helicase unwinding and that the WRN helicase can unwind a longer forked duplex without shortening of the substrate by the WRN exonuclease. Although we used the WRN proteins in which exonuclease activity was inactivated either by a mutation in the exonuclease domain or by deletion of the exonuclease domain, it is difficult to exclude the possibility that the interplay between the WRN helicase and exonuclease activities needs an intact WRN exonuclease domain, because both activities are present in the same polypeptide.

Fig. 2.

Exonuclease-deficient WRN can unwind the 2CG forked duplex without shortening the duplex. Reactions contained 1.0 nM DNA substrates and 0, 0.4, 0.8 and 1.9 ng X-WRN (lanes 1–4) or 5.5 and 11 ng Hlc-RQC (lanes 5 and 6) for the 3G and 2CG forked duplexes. Reactions were incubated at 37 °C for 15 min under the standard conditions, and reactions products were analyzed on native polyacryl-amide gels. Δ, heat-denatured substrate control.

Thus, we attempted to find a condition that reduced WRN exonuclease activity while preserving WRN helicase activity. We found that one condition containing a 1:1 M ratio of ATP to Mg² (‘exo-free’ buffer) produced the desired result. When wild-type WRN was incubated with the 3G and 2CG forked duplexes in the ‘exo-free’ buffer, no degradation of the two duplexes was detected, as shown on the denaturing gel in Fig. 3A.

Fig. 3.

WRN can unwind the 2CG forked duplex under an exonuclease activity-reduced condition. Reactions, containing 0, 0.8, and 1.5 ng WRN (lanes 1–3) and the 3G or 2G forked duplexes (1.0 nM), were and incubated at 37 °C for 15 min under the standard condition with ATP to Mg (1–1). Reaction products were run on a denaturing polyacrylamide gel to analyze exonuclease activity (A) and on a native polyacrylamide gel to analyze helicase activity (B). Δ, heat-denatured substrate control.

Next, we compared the unwinding of the 2CG and 3G forked duplexes in the ‘exo-free’ buffer. The 2CG substrate was significantly unwound without the digestion of the substrate, whereas neither unwinding nor degradation was observed with the 3G substrate (Fig. 3B). These results support no unwinding of the 3G substrate (see Fig. 2).

3.3. Single-stranded oligonucleotides used in forked duplexes can form a secondary structure

Since 2CG forked duplex was unwound in the absence of RPA and WRN exonuclease digestion, it is possible that there are other factors in the forked duplex to stimulate WRN helicase.

We investigated a DNA structural element that influences WRN helicase effectiveness because a DNA secondary structure, like a hairpin loop structure in the displaced strand, may prevent re-anne aling it to the complementary strand. To predict DNA secondary structure in the 3G and 2CG oligonucleotides, the two oligonucleo-tides were subject to the RNA fold algorithm, as described in Section 2. The prediction revealed that the 2CG oligonucleotide had the potential to form a hairpin loop structure, whereas the 3G oligonucleo-tide had no hairpin loop structure (Fig. S3A).

To see the influence of the positions of the hairpin loop structure on WRN helicase unwinding, we designed three different forked duplexes for the hairpin loop sequence, located in different positions, as shown in Fig. S3B. In sl-2, the hairpin loop sequence is positioned proximal to the fork. The hairpin loop sequence is positioned distal to the fork (sl-3). The two hairpin loop sequences are located in two positions (sl-4). The prediction showed that these oligonucleotides had the potential to form hairpin loop structures in the designed positions (Fig. S3B).

To determine the existence of a secondary structure and accuracy of these computational predictions, we utilized native polyacrylamide gel electrophoresis (Fig. 4A). We reasoned that any double-strand arrangement would alter the mobility of oligonucleotides with respect to the poly(T) counterpart that should retain the normal linear single-stranded form. 2CG, sl-2, and sl-3 single-stranded oligonucleotides migrated more quickly than the 49 poly(T), indicating that the three oligonucleotides contained an intramolecular DNA secondary structure that promotes a more rapid mobility (lanes, Fig. 4A). We used 34G and 34 poly(T), which have been reported to have hairpin loops previously, as positive controls [20]. The 34G oligonucleotide migrated more quickly than the 34 poly(T) (Fig. 4A).

Fig. 4.

The location effects of the hairpin loop on WRN helicase. (A) Secondary structure analyses of 34-bp forked duplexes. DNAs were incubated at 37 °C for 15 min in reaction buffer plus 10% glycerol. Oligonucleotides were immediately loaded on a 10% nondenaturing polyacrylamide gel. ds, forked duplex; ds/Δ, forked duplex was heat-denatured; ss, single-stranded oligonucleotide. (B) Reactions, containing the sl-2, sl-3, or sl-4 forked duplexes (1.0 nM) and 0, 3.0, 6.0, 12 ng XWRN (lanes 1–4) were incubated at 37 °C for 15 min under the ‘exo-free’ condition. Reaction products were run on a native polyacrylamide gel. Δ, heat-denatured substrate control.

To probe secondary conformation, we used T7 endonuclease I (T7 endo I) as an alternative means, a structure-selective enzyme that recognizes and cleaves hairpin looped DNA, cruciform DNA, nonperfectly matched DNA, and Holliday junctions [20,21]. A predominant nuclease bond was only detected in 2CG oligonucleotide (Fig. S4). This T7 endo I digestion supports the existence of secondary structure as predicted for the 2CG oligonucleotide.

3.4. The location of secondary structure influences WRN helicase unwinding

To determine the effect of the location of the hairpin loop on WRN helicase unwinding, the sl-2, sl-3, and sl-4 forked duplexes were incubated with increasing amounts of X-WRN in ‘exo-free’ buffer and reaction products were analyzed on native polyacryl-amide gels (Fig. 4B).

The sl-2 and sl-4 forked duplexes were significantly unwound, displacing the 49-mer-labeled strand, and no bands migrating below the 49-mer marker were observed. The sl-3 was not significantly unwound (Fig. 4B). Since the sl-2 and sl-4 were designed to have the hairpin loop sequence proximal to the fork, early formation of the hairpin loop as WRN translocates along the nucleotides may be important to stimulate unwinding by WRN helicase. The helicase reaction products were analyzed on a denaturing polyacrylamide gel. No product smaller than a 49-mer was detected in the sl-2, sl-4, or sl-3 substrates (data not shown). These results indicate that the formation of DNA secondary structure in the displaced strand behind WRN, as it translocates the fork, can stimulate unwinding by WRN helicase without shortening of the fork.

4. Discussion

Here, we found that WRN helicase can unwind a longer forked duplex, depending on a secondary conformation of the displaced ssDNA and propose that WRN helicase and exonuclease can be uncoordinated.

The slowing of WRN exonuclease digestion at GGG repeats suggests that DNA sequence context influencing the stability of duplex may regulate WRN exonuclease digestion. Thus, we increased the number of GC base pairs and changed the GGG repeat to GGGG or CGCG (Figs. 1A and S1). The 2CG forked duplex was much less digested than the 3G or 4G forked duplexes (Fig. 1A). When we examined the unwinding of these forked duplexes, interestingly, WRN unwound almost the full length of the 2CG forked duplex, as compared to the 3G or 4G forked duplexes (Fig. 1B). These data suggest that factors other than the DNA sequence may influence WRN exonuclease digestion and WRN helicase unwinding and that WRN can unwind a forked duplex without the coordinated action of two WRN activities on the fork.

Exonuclease activity-deficient full-length WRN (X-WRN) and exonuclease domain-deleted WRN fragment (HlcRQC) were able to unwind the 2CG forked duplex but were unable to unwind the 3G forked duplex (Fig. 2). Thus, these data suggest that DNA sequences in the 2CG oligomer may play a role in uncoordinated action and that WRN helicase alone can unwind the longer forked duplex.

Although WRN helicase and exonuclease activities are separable, as shown in Fig. 2, the usage of wild-type WRN would be ideal for understanding uncoordinated action instead of using exonuclease-defected WRN or the WRN fragment, because both WRN heli-case and exonuclease are parts of the same polypeptide. We found an ‘exo-free’ reaction buffer in which the molar ratio of Mg²⁺ to ATP is 1–1. Under this ‘exo-free’ buffer, neither the 3G nor the 2CG forked duplexes were digested by wild-type WRN, but only the 2CG forked duplex was mostly unwound (Fig. 3). In addition, X-WRN could efficiently unwind the 2CG forked duplex (Fig. 3). Under an equimolar ratio of Mg and ATP, Choudhary et al. reported optimal WRN helicase activity, while low-level WRN exonuclease activity on a short forked duplex (19-bp duplex) [22]. However, no observation was reported on a longer forked duplex, such as the 34-bp forked duplex. Another recent report showed a condition in which the WRN exonuclease activity was reduced on a 84-bp forked duplex by decreasing the Mg²⁺ concentration [23].

Two previously reported observations showed uncoordinated action of WRN helicase and exonuclease. WRN is able to completely unwind the 16- and 22-bp forks while WRN exonuclease is inhibited [13]. This phenomenon is explained by a rapid unwinding of the duplexes. In addition, WRN can unwind a 34-bp forked duplex in the presence of human replication protein A (hRPA), which stimulates WRN helicase to unwind the fork while inhibiting WRN exonuclease [24]. It is proposed that the interaction of hRPA with WRN and the binding of hRPA to the displaced ssDNA may prevent re-annealing of the unwound strand as WRN translocates.

We, therefore, hypothesized that the DNA secondary structure in the being displaced strand during unwinding may play a role in preventing reannealing of the unwound portion to its complementary strand, thereby stimulating WRN helicase to unwind forked duplexes. In fact, the 2CG 49-mer was predicted to have a putative hairpin loop structure shown in Fig. S3A and had a more rapid electrophoretic mobility shift relative to the corresponding the 49-mer poly(T)-DNA (Fig. 4A). Further, T7 endo I digesting analysis indicates that the being displaced strand can form an intramolecular DNA secondary structure (Fig. S4).

Since the 2CG oligonucleotides have the potential to form secondary structures, we tested whether the 2CG forked duplex generated by annealing with the 2CG oligonucleotides contains a normal DNA double helix. We analyzed the 2CG forked duplex with a restriction enzyme, Afl III, because the forked duplex contains two restriction sites of Afl III (A/CryGT). Two bands, corresponding to the predicted products (21 nts and 28 nts) of the Afl III digestion, were detected (Fig. S2). This digestion pattern indicates that the 2CG fork consists of a normal DNA double helix.

To prevent re-annealing of the displaced strand as helicase translocates, the timing of formation of DNA secondary structure may be crucial. If a DNA secondary structure was formed before WRN dissociates during unwinding, then WRN can easily recognize the remaining fork and continue to unwind. However, if the displaced strand reannealed during WRN unwinding or after WRN dissociation, it may not be feasible for WRN to find a proper site for unwinding, such as an ss/ds junction. The sl-2 and sl-4 forked duplexes, containing a hairpin loop proximal to the fork, are more favorable than the sl-3 for WRN helicase unwinding (Fig. 4B). This result implies that formation of a hairpin loop on the displaced strand just upon being unwound may prevent re-annealing the displaced strand to its complementary strand, thus allowing WRN to translocate effectively.

This study is the first to show that the formation of DNA secondary conformation in a displaced strand can stimulate WRN helicase activity while inhibiting WRN exonuclease activity. The DNA secondary structure may affect WRN helicase actions through the following processes.

WRN can dissociate from DNA substrates with some finite probability during the unwinding process due to Low unwinding processivity of the WRN helicase. Once dissociated from the WRN, any partially unwound DNA intermediate will reanneal to reform the original duplex. The DNA secondary structure in the displaced strand of the partially unwound DNA intermediate may prevent reannealing, consequently promoting reinitiation of the DNA intermediate unwinding.

The ssDNA annealing activity of WRN may lead to reanneal the partially unwound DNA intermediate. This activity may be detected as a repetitive unwinding. A single-molecule analysis showed that BLM RecQ helicase unwound individual DNA molecules in a repetitive manner [25]. Thus, a single-molecule analysis of WRN in either the absence or the presence of RPA would be useful to test repetitive unwinding of WRN. If a repetitive unwinding is observed even in the presence of RPA, that indicates that RPA may not generally block reannealing. If a repetitive unwinding is not observed, other single strand binding proteins such as Escherichia coli SSB should stimulate WRN helicase activity. Instead of using single strand binding proteins, the substrate used in this study may be very useful for a single-molecule analysis of WRN. If no repetitive unwinding is observed, a secondary structure formed in the unwound strand may block reannealing activity of WRN.

The intrinsic barrier of the low processivity of WRN would be overcome by the formation of DNA secondary structure. Thus, WRN may be able to unwind longer duplex DNA. Since other heli-cases, such as BLM, RecQ1, and FACNJ, are also stimulated by hRPA [26,27], it remains to be investigated whether DNA secondary formation replaces the requirement of hRPA for unwinding of forked duplexes by these enzymes.

Forked duplexes used in vitro regarding the coordinate action of WRN activities seem to be very limited in resembling in vivo DNA structures because in vitro forked duplexes are DNA segments which consist of a blunt end at one end and a fork at the other. In vivo DNA structures that occur as intermediates during in vivo processes, such as replication, repair, and recombination, have various sizes and structures unlike a simple forked structure. Thus, the forked duplexes raise several interesting questions. How are forked duplexes generated from chromosomal DNA in vivo? How large are the forked duplexes? How biologically relevant is it for helicase to unwind the forked duplexes?

Abbreviations

RPA

replication protein A

WRN

Werner syndrome protein

WS

Werner syndrome

dsDNA

double-stranded DNA