Modulation of Werner Syndrome Protein Function by a Single Mutation in the Conserved RecQ Domain*

Naturally occurring mutations in the human RECQ3 gene result in truncated Werner protein (WRN) and manifest as a rare premature aging disorder, Werner syndrome. Cellular and biochemical studies suggest a multifaceted role of WRN in DNA replication, DNA repair, recombination, and telomere maintenance. The RecQ C-terminal (RQC) domain of WRN was determined previously to be the major site of interaction for DNA and proteins. By using site-directed mutagenesis in the WRN RQC domain, we determined which amino acids might be playing a critical role in WRN function. A site-directed mutation at Lys-1016 significantly decreased WRN binding to fork or bubble DNA substrates. Moreover, the Lys-1016 mutation markedly reduced WRN helicase activity on fork, D-loop, and Holliday junction substrates in addition to reducing significantly the ability of WRN to stimulate FEN-1 incision activities. Thus, DNA binding mediated by the RQC domain is crucial for WRN helicase and its coordinated functions. Our nuclear magnetic resonance data on the three-dimensional structure of the wild-type RQC and Lys-1016 mutant proteins display a remarkable similarity in their structures.

helicases based on cellular and biochemical evidence suggest their involvement in DNA replication, DNA repair, recombination, and telomere maintenance. Clinically, defects in certain RECQ genes have been linked to rare genetic disorders, including Bloom syndrome (RECQ2), Werner syndrome (RECQ3), and Rothmund-Thompson syndrome (RECQ4). One of the prominent characteristics common to all RecQ helicases is the sequence conservation of the seven helicase motifs centrally located in the respective proteins (12). In addition to the helicase domain, a majority of RecQ helicases shares another conserved sequence designated the RecQ C-terminal (RQC) region (13). The WRN RQC region is located between amino acid (aa) residues 872 and 1045 and is homologous to the part of the catalytic core of Escherichia coli RecQ for which the three-dimensional structure has revealed the presence of a winged helix-turn-helix (wHTH) motif (14). Moreover, of the 17 WRN-interacting proteins, 14 that are functionally significant require aa 949 -1092 or at least part of this region for the interaction of WRN with them (15).
The wHTH motif in mammalian proteins has been demonstrated to be important for protein-DNA interactions as well as protein-protein interactions (16 -20). The wHTH motif mediates DNA-protein interaction through the "DNA recognition helix" of the helix core consisting of three ␣-helices. In general, the recognition helix interacts with the major groove of DNA as shown in the DNA-protein co-crystal structures of a variety of proteins involved in DNA metabolism (17)(18)(19). Therefore, the presence of the wHTH motif in the catalytic core of E. coli RecQ and the possibility that the wHTH motif may also exist in the WRN RQC domain offer some explanation as to why the WRN RQC fragment (aa 949 -1092) is the strongest DNA binding region followed by the helicase RNaseD C-terminal (HRDC) domain and the exonuclease domain (21). The versatility of WRN to bind and act upon various DNA substrates in vitro such as dsDNA flanked by a 3Ј ssDNA tail (6,22), synthetic replication fork (23), Holliday junction (21,24), bubble (25), triple helix (26), G4 tetraplex (27), and telomeric D-loop (28) suggest that WRN may harbor a structure-specific DNA-binding motif, such as wHTH in the RQC domain, in addition to auxiliary DNAinteracting surfaces such as HRDC and exonuclease domains. Therefore, in order to understand the importance of the RQC domain in the DNA binding mechanism of WRN, site-directed mutagenesis was used to introduce missense mutations in the RQC domain of WRN to identify the critical aa mediating the WRN-DNA interaction. After identifying the critical aa that modulates the DNA binding activity of WRN, we demonstrated how this mutation affects the WRN helicase activity and the FEN-1 incision activity. Furthermore, we proved that the modulations of the WRN-coordinated functions were not a direct consequence of the altered three-dimensional structure of WRN via solution nuclear magnetic resonance.

EXPERIMENTAL PROCEDURES
Mutagenesis of WRN-Site-directed mutagenesis was performed in order to change the wild-type aa residues, Lys-1008, Gln-1010, Lys-1016, and Arg-1020, to alanine in the region of WRN RQC using the QuikChange II site-directed mutagenesis kit (Stratagene). The sequences of primers used for site-directed mutagenesis are shown in TABLE ONE. PCR was performed using pGEX-KG-WRN-(949 -1092) as a template, and PCR products were sequenced as described previously (21).
Recombinant Proteins-All of E. coli WRN wild-type or mutant fragments were cloned into Gateway pDEST 15 (Invitrogen) according to the manufacturer's guidelines or pGEX-KG vectors, and the recombinant glutathione S-transferase (GST)-tagged proteins were purified as described previously (21). His-tagged recombinant flap endonuclease-1 (FEN-1) was prepared as described previously (29). Protein concentrations were determined by using the following methods: Bradford assay (Bio-Rad), Silver Xpress staining (Invitrogen), Amido Black staining, and GelCode Blue staining (Pierce). Western blot analysis was also used to confirm the relative concentration of each protein using mouse anti-GST antibody and rabbit anti-mouse horseradish peroxidase-conjugated antibody.
DNA Substrate Preparation-The fork substrates that have 19-and 34-bp duplex DNA with 3Ј and 5Ј single-stranded overhangs were prepared as described previously (23,30). The bubble substrates containing a 12-nt bubble and the Holliday junction containing a 12-bp homologous core were prepared as described previously (25). The 1-nt 5Ј flap substrate was prepared using Tstem25 and FLAP26, and D-loop substrates were prepared as described previously (31)(32)(33).
Electrophoretic Mobility Shift Assay-EMSAs were performed as described previously (21) with modifications. 20-l reactions contained 30 fmol of DNA substrates in a buffer containing 40 mM Tris (pH 7.0), 1 mM EDTA, 20 mM NaCl, 20 g/ml bovine serum albumin (BSA), and 8% glycerol. As indicated in the figure legends, 5 ng of poly(dI-dC) was preincubated with WRN fragment on ice. The reaction mixtures were incubated on ice for 30 min on ice and electrophoresed in 5% native polyacrylamide gel at 4°C. The electrophoresis was carried out in 1ϫ TAE (40 mM Tris acetate, 1 mM EDTA (pH 8.0)) at constant voltage of 12.8 V/cm for 2 h, and the resulting band-shifts were visualized using PhosphorImager and ImageQuant software (Amersham Biosciences).
Enzyme-linked Immunosorbent Assay-Purified recombinant FEN-1 protein was diluted to 1 ng/l in carbonate buffer (0.016 M Na 2 CO 3 , 0.034 M NaHCO 3 (pH 9.6)) and added to the appropriate wells of a 96-well microtiter plate (50 l/well). The plate was incubated at 4°C overnight. The wells were washed with 1ϫ phosphate-buffered saline (PBS). For blocking, 200 l of blocking buffer (1ϫ PBS, 0.5% Tween 20, and 3% BSA) was added to each well and incubated for 1 h at 37°C. After washing twice with 300 l of 1ϫ PBS, 50 l of either purified recombinant GST RQC or K1016A mutant fragments was applied to each well for binding to FEN-1 at various concentrations, and samples were incubated at 37°C for 1 h. Purified recombinant GST protein was used as a control for RQC or K1016A mutant. BSA was also used as a control with RQC, K1016A mutant, or BSA alone to measure nonspecific binding. For ethidium bromide (EtBr) treatment, 50 g/ml EtBr was included in the incubation with FEN-1 and RQC or K1016A mutant during the incubation for binding. After the incubation, the wells were washed five times with 1ϫ PBS. Anti-GST mouse antibody (Santa Cruz Biotechnology) was added at 1:500 in blocking buffer, and the samples were incubated at 37°C for 1 h. The wells were washed five times and incubated with anti-mouse horseradish peroxidase-conjugated secondary antibody (1:5000) in blocking buffer for 1 h at 37°C. After washing five times, 50 l of developing buffer (0.05 M phosphate/citrate buffer (pH 5.0), 0.06% H 2 O 2 ) containing o-phenylenediamine dihydrochloride (1 tablet/10 ml of developing buffer; Sigma) was added. The reaction was terminated after 5 min by adding 25 l of 3 M H 2 SO 4 , and the absorbance was measured at 490 nm. The fraction of RQC or K1016A mutant specifically bound to FEN-1 protein was determined from the enzymelinked immunosorbent assay, and the Hill plot was used for the calculation of K d values as described previously (34).
Flap Endonuclease-1 Incision Assay-FEN-1 incision assay was performed as described previously (29) with modifications. 20-l reactions contained 10 fmol of DNA substrate, 8 fmol of FEN-1, and the indicated amounts of RQC or the Lys-1016 mutant in a buffer containing 150 mM HEPES (pH 7.5), 200 mM KCl, 40 mM MgCl 2 , 500 g/ml BSA, and 25% glycerol. The RQC or the Lys-1016 mutant was mixed with the substrate and buffer on ice prior to the addition of FEN-1. Reactions were incubated at 37°C for 15 min and terminated with 10 l of stop solution (80% formamide (v/v), 0.1% bromphenol blue, and 0.1% xylene cyanol). After heating reaction mixtures to 95°C for 5 min, they were electrophoresed in 20% polyacrylamide, 7 M urea denaturing gels.
Helicase Assay-The reaction conditions for helicase assays and time course experiments were performed as described previously (35). 20-l reactions contained 30 mM HEPES (pH 7.4), 5% glycerol, 40 mM KCl, 100 ng/l BSA, 0.8 nM DNA substrate, 2 mM MgCl 2 , and 2 mM ATP. Reactions were incubated for 15 min (unless otherwise indicated) at 37°C and terminated using 20 l of stop buffer (35 mM EDTA, 0.6% SDS, 25% glycerol, 0.04% bromphenol blue, 0.04% xylene cyanol) containing a 10-fold excess of unlabeled oligonucleotide of the same sequence as the labeled strand of the DNA fork substrate. Reaction mixtures were electrophoresed in nondenaturing 12% polyacrylamide gels, and the results were analyzed using PhosphorImager and Image-Quant software (Amersham Biosciences).
NMR Sample Preparation-The cloning and expression of RQC motif of the WRN protein for NMR studies have been described previously (36). Briefly, the plasmids encoding the wild-type and the K1016A mutant WRN protein were used as templates to subclone the KpnI and XhoI fragment containing the 144-residue RQC domain of WRN into pET32d, the mutant pET32a (Novagen) in which the GGTATG sequence right after the thrombin cleavage site was mutated to introduce a KpnI site. These plasmids, pET32d-RQC, were transformed into the E. coli strain BL21 (Novagen), grown in M9 minimal medium containing either 15

Sequences of primers used for site-directed mutagenesis
Underlines indicate the locations of mutation to alanine. NMR Spectroscopy-NMR experiments were recorded at 22°C on a Bruker DRX-600 spectrometer equipped with z-shielded gradient triple-resonance cryoprobe. All data were processed with the NMRPipe package (37) and analyzed using the PIPP, CAPP, and STAPP programs (38). The sequential backbone assignments of the RQC domain were made using triple-resonance methodology (39). The two-dimensional spin-echo difference experiments, HNCG_arom and HN(CO)CG_arom, were used to identify 1 H N -15 N resonances of aromatic residues and the following residues (40), which were used as starting points for automatic sequential backbone resonance ( 1 H N , 15 N, 13 C ␣ , and 13 C ␤ ) assignments when combined with three-dimensional HNCACB, CBCA-(CO)NH, and HNCA experiments. Three-dimensional HNCO, HNHA, and HNCOCO (41) experiments were performed to obtain 1 H ␣ and carbonyl ( 13 CЈ) resonance assignments and to simultaneously obtain restraints. 1 H ␣ and 1 H ␤ resonances for the majority of residues were obtained by using the three-dimensional HBHA(CO)NH experiment. The two-dimensional HNCG_arom and HN(CO)CG_arom were used to determine intra-residue 3 J NC␥ and 3 J CЈ C␥ coupling constants (40).
Transfection and Immunofluorescence Assays-SV40-transformed human kidney epithelial (293T) cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) and 10% fetal bovine serum. For expression as GFP fusion protein in mammalian cells, cells growing on coverslips were transfected with wild-type pEGFPC3-WRN (aa 1-1432) or pEGFPC3-WRN-K1016A mutant (aa 1-1432) containing a single missense mutation at Lys-1016 using the PolyFect transfection reagent (Qiagen). After 40 h, the cells were fixed and permeabilized with 3.5% paraformaldehyde (Sigma), 0.2% Triton X-100 (Sigma) in 1ϫ PBS for 15 min each at room temperature. Then the coverslips were washed three times with PBS followed by incubation with blocking buffer (0.1% Tween 20, 2% BSA in 1ϫ PBS) for 1 h at room temperature. Finally, coverslips were mounted on Vectashield containing DAPI (Vector Laboratories) and viewed under a AxioVision 3.1 microscope (Zeiss) with deconvolution (42).

A Putative wHTH Motif in the RQC Domain of WRN-
The threedimensional crystal structure of E. coli RecQ helicase has revealed the presence of a wHTH motif (14). Thus, we hypothesized that WRN may also contain a wHTH motif responsible for mediating the major DNA binding activity of WRN. In support of a similar DNA-binding motif in WRN RQC, secondary structure predictions revealed a helix motif in the WRN RQC that directly matched the putative DNA recognition helix of E. coli RecQ (supplement 1). Sequence homologies between the putative DNA recognition helix in WRN and the other RecQ helicases were examined using multiple-sequence alignment software, ClustalW, and four amino acids were selected for site-directed mutagenesis ( Fig. 1, arrows). Conserved as a positively charged amino acid in 5 of 7 RecQ helicases, the WRN Lys-1008 located outside the putative DNA recognition helix was chosen to serve as a control to missense mutations made inside the putative DNA recognition helix. Likewise, the polar Gln-1010 located outside the putative DNA recognition helix was chosen as a second control. Inside the putative DNA recognition helix of WRN RQC, we found only two positively charged aa residues, Lys-1016 and Arg-1020, which are likely to be involved in protein-DNA interactions. Significantly, these residues are conserved as a positive charged aa in 6 of 7 and 4 of 7 RecQ helicases, respectively. After the confirmation of a single missense mutation at each four aa residues, the wild-type RQC and RQC mutant proteins were prepared. The R1020A mutant could not be purified to homogeneity as other WRN mutant fragments and was excluded from our studies (data not shown).
Lys-1016 of WRN Plays a Significant Role in Binding Fork and Bubble DNA Substrates-To compare relative DNA binding affinities of these mutant RQC proteins, it was critical to use precisely the same amount of each protein. Therefore, each protein preparation and the relative amount of each protein fragment were analyzed carefully, and the quantitative results were in close agreement using several different methods: Bradford assay, Amido Black staining, GelCode Blue staining ( Fig. 2A),  (14) is marked above the sequence alignment of RecQ helicases with significant sequence homologies to the WRN RQC domain. Arrows indicate the locations of site-directed mutagenesis: Lys-1008, Gln-1010, Lys-1016, and Arg-1020 from left to right. The accession number for each protein is indicated next to the name of the protein. Gray boxes indicate some aa residues carrying either a similar charge at the physiological pH or similar hydrophobicity. The symbol "ϳ" indicates the only place that was modified by hand in the sequence alignment. The numbers above each protein fragment, RQC, Hlc, and HlcRQC, indicate the aa boundaries. NOVEMBER 25, 2005 • VOLUME 280 • NUMBER 47 and silver staining (Fig. 2C). In addition, Western blot analysis was performed to verify the relative concentrations of wild-type RQC, K1008A, Q1010A, and K1016A (Fig. 2B) and Hlc, HlcRQC, and Hlc-K1016A proteins (Fig. 2D).

Mutation in WRN RQC Modulates Protein Function
Because the RQC domain was demonstrated to be the strongest DNA binding region of WRN (21), we examined which of the three amino acids, Lys-1008, Gln-1010, or Lys-1016, was important for the WRN-DNA interaction. The wild-type RQC and mutants K1008A, Q1010A, and K1016A were used to perform EMSAs with various DNA substrates, including the fork, bubble, D-loop, or the Holliday junction substrates. These substrates were selected to reflect the types of DNA structures that WRN might be expected to encounter in vivo (21). When radiolabeled fork substrate was used, both the wild-type RQC and Q1010A mutant showed nearly identical concentration-dependent DNA mobility shifts (Fig. 3A). The EMSA results using the K1008A mutant generated band-shift patterns similar to that of the wild-type RQC (data not shown). However, when the K1016A mutant was used with the same fork DNA substrate, minimal band-shifts were observed, suggesting that Lys-1016 plays a significant role in the WRN-DNA interaction (Fig. 3B).
By using a bubble DNA substrate, the K1016A mutant displayed significantly reduced DNA binding, whereas the K1008A mutant or the Q1010A mutant generated band-shifts similar to the wild-type RQC (Fig. 3, C and D; K1008A mutant data not shown). Each of these experiments was performed multiple times, and band-shifts obtained from these independent experiments were quantified by using ImageQuant and plotted with standard deviations as shown in supplemental 2A (fork substrate) and 2B (bubble substrate).
A Single Missense Mutation at Lys-1016 of WRN Results in Decreased Stimulation of FEN-1 Incision Activity-We showed previously that WRN RQC (aa 949 -1092) strongly stimulates the incision activity of FEN-1 in vitro (33). To investigate how the WRN-DNA interaction might affect the functional stimulation of FEN-1 cleavage, incision assays were performed by using 1-nt flap substrates in the presence of wild-type RQC or the K1016A mutant (Fig. 4A). Wild-type RQC stimulated FEN-1 incision activity (Fig. 4A, lanes 5-7) similar to previous results (33). However, there was little or no stimulation of FEN-1 incision activities using the K1016A mutant (Fig. 4A, lanes 8 -10). The incised products from two independent experiments were quantified and shown below the gel picture. From the observations that the K1016A mutant had decreased affinities for fork and bubble substrates (Fig. 3, A-D), we hypothesized that the lack of FEN-1 stimulation by the K1016A mutant might be because of a decreased affinity of the K1016A mutant for the 5Ј flap substrate. However, it was also possible that introduction of a K1016A mutation altered the WRN-FEN-1 protein-protein interface, resulting in a reduced ability of the WRN mutant to stimulate FEN-1 cleavage.
To address the impact of the K1016A mutation on the ability of the RQC to bind FEN-1, enzyme-linked immunosorbent assays were performed. Recombinant FEN-1 protein was adsorbed onto a polystyrene surface of a 96-well plate via hydrophobic interactions and was allowed to bind to either the K1016A mutant or wild-type RQC. FEN-1 binding by wild-type RQC and the K1016A mutant was very similar (Fig. 4B). K d values for the interaction of FEN-1 with RQC or with the WRN mutant were 9.50 and 9.58 nM respectively. This suggested that the introduction of K1016A mutation did not affect the protein-protein interaction between FEN-1 and the WRN mutant.
Next, we wanted to determine explicitly whether the decreased stimulatory effect of the K1016A mutant on the FEN-1 activity was because of decreased DNA binding activity. Therefore, EMSA was performed using the same flap substrate used in FEN-1 incision assays. As shown in Fig. 4C, the K1016A mutation severely reduced the band-shifts as compared with the wild-type RQC or the Q1010A mutant. The amount of band-shifts from two independent experiments was quantified and plotted with error bars in supplement 2C. In summary, our data suggest that the mutation at Lys-1016 drastically reduced the ability of WRN RQC to interact with the 5Ј flap DNA substrates. Thus, this mutant affected the functional stimulation of FEN-1 cleavage but not the physical interaction between WRN RQC and FEN-1.
Functional Analysis of Mutant WRN K1016A Helicase Activity-Next, we investigated whether the helicase activity of WRN was modulated by the presence or absence of the DNA binding domain (RQC) and by the mutation at Lys-1016 in the putative wHTH motif. Three recombinant WRN fragments, Hlc, HlcRQC, and the Hlc-K1016A mutant, were purified (Fig. 2, C and D). Hlc contains the helicase domain and a small N-terminal fragment of the RQC domain but lacks the putative wHTH motif. HlcRQC contains the helicase domain and the wild-type RQC. Hlc-K1016A mutant contains the helicase domain and the RQC with the K1016A mutation in the putative wHTH motif. Radiometric helicase assays were performed by using preferred DNA substrates of WRN including the fork, D-loop, and the Holliday junction substrates. As shown in Fig. 5A, the unwinding of the 19-bp fork substrate was greatest with HlcRQC. When the Hlc-K1016A mutant was used, the  Fig. 2A. The anti-GST mouse antibody and anti-mouse horseradish peroxidase-conjugated rabbit antibody were used with chemiluminescence detection kit. Their sizes correspond to ϳ43 kDa, including GST tag. C, HlcRQC and Hlc-K1016A mutant proteins (300 ng each, left) and Hlc (100 ng, right) were stained with silver staining. D, Western blot analysis using 0.47 pmol of proteins shown in C. Apparent molecular mass for HlcRQC and Hlc-K1016A mutant corresponds to ϳ93 kDa, and Hlc corresponds to ϳ76 kDa, including GST tag.
helicase activity was distinctly reduced. A comparison of unwinding activities between HlcRQC and Hlc-K1016A mutant at 0.75 nM protein concentration showed that HlcRQC could unwind ϳ90% of the labeled fork, whereas the Hlc-K1016A mutant unwound ϳ40% of the substrate (Fig. 5A, lanes 6 and 12). Furthermore, the unwinding activity of Hlc, lacking the putative wHTH motif, was severely reduced. Even at the highest concentration (12 nM), only ϳ25% of the fork substrate was unwound, suggesting a requirement of the RQC domain for optimal activity of WRN helicase (Fig. 5A, lane 16).
We next performed time course assays to further substantiate the results from the helicase assays. Over 70% of the 19-bp fork substrate was unwound by HlcRQC compared with 35% unwinding by Hlc-K1016A mutant after 16 min of incubation time at 0.75 nM (Fig. 5B). When the concentration of the Hlc-K1016A mutant was increased to 1.5 nM, we observed an increase in DNA unwinding activity to ϳ66% approaching that of the wild-type HlcRQC after 16 min. On the other hand, there was a minimal unwinding activity by Hlc, and it did not increase even at 16 min (Fig. 5B). These results correlate with radiometric helicase assays and DNA binding studies as shown in Fig. 5A and supplement 2D, respectively.
We subsequently investigated D-loop and Holliday junction DNA substrates that have been shown to be preferred substrates for WRN in vitro (33,39) and most likely also in vivo during homologous recombination using the radiometric helicase and time course assays. The results from assays using the D-loop substrate were similar to the results obtained using the 19-bp fork substrate. HlcRQC unwound over 90% of the D-loop substrate compared with the Hlc-K1016A mutant which unwound ϳ45% after 16 min of incubation time (Fig. 5C). The unwind-ing activity of the Hlc remained below 4% throughout the entire 16-min incubation time. Also, the examination of Holliday junction substrates in these assays showed analogous results to the 19-bp fork and D-loop, such that the helicase activity was HlcRQC Ͼ Hlc-K1016A mutant Ͼ Hlc (data not shown). These results strongly support the role of the RQC domain and Lys-1016 residue in WRN helicase function.
To evaluate if the reduction of WRN helicase activity in Hlc-K1016A or Hlc reflected a decrease in DNA binding affinity, EMSA was performed using the 19-bp fork substrate. As shown in Fig. 5D, there was less than 9% band-shift when the Hlc lacking the RQC region was used. This can be correlated to the negligible helicase activity of Hlc. Also, the HlcRQC exhibited over 85% band-shift in lanes 3-5 compared with ϳ45% band-shift of the Hlc-K1016A mutant in lanes 8 -10, indicating the importance of Lys-1016 for DNA binding activity, which again can be correlated directly to the helicase activities of each respective protein fragment (Fig. 5, A-C, compare with D). The band-shifts from two independent experiments were quantified and plotted with error bars as shown in supplement 2D. The same experiments were also performed using the Holliday junction substrate. The results obtained were qualitatively analogous to those of the 19-bp fork substrates as discussed above (data not shown).
Structural Comparison between Wild-type RQC and the K1016A Mutant Proteins by NMR Spectroscopy-We found that a single mutation at Lys-1016 in the RQC domain, but not at Lys-1008 and Gln-1010, profoundly alters fundamental functions of WRN RQC, i.e. its DNA binding/helicase activity and the stimulation of FEN-1 incision activity. An important question was whether the K1016A mutation alters the three-dimensional structure of the WRN RQC domain. In order to eval-  NOVEMBER 25, 2005 • VOLUME 280 • NUMBER 47 uate a K1016A mutation-induced conformational change in the threedimensional structure of the RQC domain, we compared chemical shift differences between the wild-type RQC and the K1016A mutant proteins by NMR spectroscopy. Solution NMR spectroscopy provides structural information at atomic resolution about macromolecules, and NMR chemical shifts are sensitive to both the local environment and the global structure of the protein. Recently, the NMR resonance assignments of the RQC motif of WRN were reported (36). In the present study, we have analyzed both the wild-type RQC and the K1016A mutant proteins in order to compare their structural differences. Essentially complete backbone resonances ( 1 H N , 15 N, 13 C ␣ , and 13 CЈ) were assigned for 144-aa residues of both the wild-type and K1016A proteins using double and triple resonance multidimensional heteronuclear NMR spectroscopy, making use of uniformly 15 N/ 13 C-labeled proteins (36,39). Fig. 6 shows the superimposed 1 H N -15 N heteronuclear single quantum coherence (HSQC) spectra of wild-type RQC (red) and the K1016A mutant (black). The results provide direct evidence that both the wild-type RQC and the K1016A mutant protein have very similar structures. Fig. 7 shows the weighted average of the backbone amide 1 H N and 15 N chemical shift differences between two proteins. For most of the aa residues, this average is considerably smaller than 0.1 ppm, with the exception of the mutation site Lys-1016, which has the largest chemical shift difference of 0.336 ppm. The chemical shift differences for three residues adjacent to the mutation site, i.e. Glu-1012, Ser-1013, and Ala-1017, are 0.073, 0.085, and 0.165 ppm, respectively, indicating that the K1016A mutant protein retains a remarkably similar structure as the wild-type RQC protein with some subtle structural differences close to the mutation site.

Mutation in WRN RQC Modulates Protein Function
Both Full-length Wild-type WRN and WRN-K1016A Mutant Proteins Exhibit Similar Subnuclear Localization Patterns-To address whether the full-length WRN-K1016A mutant exhibits different subnuclear localization patterns compared with the wild-type WRN because of its compromised ability to interact with DNA, we transfected full-length wild-type WRN or WRN-K1016A mutant tagged with green fluorescent protein (GFP) in 293T cells. In Fig. 8, the merged picture showing both DAPI staining and green fluorescence indicates that both wildtype WRN and the WRN-K1016A mutant partially localize to regions of poor DAPI staining within the nucleus. Previously, the RNA-rich regions have been shown to stain poorly using DAPI because of the selectivity of DAPI for DNA over RNA (51). Thus, DAPI-poor staining regions have been understood as regions of nucleoli that are active sites of transcriptions in the nucleus. In addition, similar experiments were performed that included nucleoli staining by using an anti-nucleolin antibody (data not shown), because the wild-type WRN protein has been characterized previously to localize in regions of nucleoli (43). These experiments showed that there is no significant difference in the wild-type and the WRN-K1016A mutant in their abilities to localize  partially to regions of nucleoli. Therefore, this suggests that the mutation at Lys-1016 in WRN does not significantly change the subnuclear localization patterns observed with the wild-type WRN. Furthermore, these results demonstrated that the mutation at K1016A does not adversely affect the nuclear localization signal located in the C-terminal region of WRN (43). The nuclear localization signal is properly recognized by the nuclear membrane transporters that suggest that the WRN-K1016A mutant protein is folded properly.

DISCUSSION
The multifaceted role of the RQC domain in the overall function of WRN protein has been explored from several different aspects. The RQC domain is a necessary component for directing WRN proteins to the nucleoli (43). Numerous functionally significant protein-protein interactions, in addition to structure-specific interactions with DNA substrates, require the RQC domain (15,21,43). Our studies demonstrate that the Lys-1016 residue in the RQC domain is one of the critical mediators of WRN-DNA interactions. Replacement of the Lys-1016 residue with a neutral alanine maintained the protein structure based on NMR studies but severely affected the DNA binding and unwinding activities of various WRN fragments. We observed a significant decrease in DNA binding affinities of WRN fragments with the Lys-1016 mutation to fork, bubble, and flap DNA substrates. Furthermore, the K1016A mutation markedly reduced DNA unwinding activity of WRN-(500 -1104) on 19-bp fork, D-loop, and Holliday junction DNA substrates, suggesting that the stability of WRN-DNA interactions directly modulates the helicase activity of WRN fragments. In addition, the ability of the WRN-(949 -1092) to stimulate FEN-1 incision activity was prevented by the K1016A mutation. This suggests that proper DNA binding of WRN may influence the catalytic activities of other protein binding partners besides FEN-1. Our NMR data indicate that both the wild-type RQC and the K1016A mutant proteins have essentially identical three-dimensional structures and argue against the idea that the K1016A mutant may have interacted with FEN-1 in a way as to alter the conformation of FEN-1, thereby inhibiting its basal incision activities.
The importance of positively charged aa residues in protein-DNA interactions has been established previously (44,45). Furthermore, DNA-protein co-crystal structures of proteins containing a wHTH motif demonstrate the importance of positively charged aa residues in the DNA recognition helix for DNA interactions (20). All of these stud-  NOVEMBER 25, 2005 • VOLUME 280 • NUMBER 47 ies support the significance of positively charged aa residues for DNA binding and strengthen our conclusion that Lys-1016 plays a crucial role in WRN-DNA interactions. In support of this notion, the WRN RQC domain has been determined previously to be the strongest DNA binding region of WRN in vitro followed by the helicase and HRDC domains (21). We also performed a ClustalW-assisted sequence alignment of other RecQ helicases (Fig. 1). Although 6 of 7 RecQ helicases that pos-sess homologous sequence in the RQC domain contained a positively charged residue at WRN Lys-1016, the sequence homology of RTS and RecQL5 with the WRN RQC domain was poor (data not shown). This suggests that the tertiary structure of RTS and RecQL5 might be different at least in the region of the WRN RQC domain. Taken together, it is difficult to propose a general requirement for Lys-1016 in the DNA interaction of all RecQ helicases, but we argue that the positively charged residues in RecQ helicases that possess good sequence homology with the WRN RQC domain have a strong potential for mediating DNA interactions (Fig. 1). Also, recent structural characterization of the E. coli HRDC domain demonstrated a role of the HRDC domain in ssDNA binding (46). However, the WRN HRDC domain does not bind ssDNA very well (21). Therefore, the HRDC domain may govern the DNA substrate specificity of RecQ helicases and diversify each RecQ helicase to a specific cellular role in conjunction with DNA-interacting aa residues in the RQC domain.

Mutation in WRN RQC Modulates Protein Function
Our NMR studies also provided direct evidence that the K1016A mutation-induced structural change is remarkably subtle and highly localized. To examine further the structural differences in the regions close to the mutation, we selected three aromatic residues, Trp-1014, Trp-1015, and Phe-1018, in the vicinity of the Lys-1016. For the wildtype RQC protein, 3 J NC␥ couplings for Trp-1014 and Phe-1018 residues were determined to be 2.4 Hz, indicative of 1 ϭ 180°; whereas the 3 J NC␥ and 3 J CЈC␥ couplings for Trp-1015 were measured to be Յ0.5 and Յ1.1 Hz, respectively, indicative of 1 ϭ 60°. The identical 3 J NC␥ and 3 J CЈC␥ coupling constants were measured for these aromatic residues in the K1016A mutant protein, suggesting that even those residues adjacent to the Lys-1016 mutation site maintain the same side chain orientations. This conclusion is further supported by the observation that the nuclear Overhauser enhancement patterns for both wild-type and the mutant proteins were unchanged (data not shown).
Furthermore, our results suggest that the Lys-1016 mutant interacts with the DNA through its positively charged side chain located on the surface of the protein. If we assume that the amino group of Lys-1016 is located on the protein surface, the polar Gln-1010 is the 6th aa residue from the Lys-1016 with its side chain projecting ϳ120°away from Lys-1016 looking down the helical axis. Thus, the location of aa Gln-1010 may be less favorable to be exposed on the surface of the protein in the ␣-helical structure, and as a consequence, Gln-1010 may not play a direct role in the WRN-DNA interaction. Our EMSA results support this notion showing no change in DNA binding affinities using the Q1010A mutant compared with the wild-type RQC. Moreover, there is another positively charged aa Arg-1020 within the putative DNA recognition helix. It is interesting to note that Arg-1020 is the 4th aa residue from Lys-1016, located only 40°away from the position of Lys-1016 looking down the helical axis. This suggests that the location of Arg-1020 is most likely also on the surface of the protein. Although it is possible that mutations at other positively charged amino acid such as Arg-1020 within the putative DNA recognition helix could contribute to additional WRN-DNA interactions, this could not be addressed at this time due to difficulties in purifying the protein mutated at Arg-1020. Consistent with this notion, our EMSA results demonstrated that the K1016A mutant significantly decreased, but did not completely abolish, the DNA binding activities suggesting other aa residue(s) within the RQC domain must contribute to the stability of the WRN-DNA interactions.
Consistent with the previous observations of von Kobbe et al. (21), who reported a weak DNA binding activity for the Hlc fragment (aa 500 -946), we found that the HlcRQC (aa 500 -1104) containing the putative wHTH motif could bind DNA substrates with much higher affinity than the Hlc fragment (aa 500 -946). Although very weak, the fact that Hlc (aa 500 -946) could still bind to DNA substrates raises an interesting question because the x-ray crystal structure of E. coli RecQ revealed the presence of a C 4 zinc finger motif, which is homologous to aa 909 -939 residues of WRN (14). Thus, the purpose of two putative DNA-binding motifs in WRN, C 4 zinc finger motif in aa 909 -939 and wHTH motif in aa 949 -1092, might be for a stronger DNA binding anchor such as the RQC domain to hold onto the upstream duplex DNA, whereas the helicase domain unwinds at the fork DNA junction. In support of this notion, when we added the RQC fragment and the Hlc fragment separately in the helicase reaction mixture, we observed minimal helicase activity, suggesting a requirement for a mechanical connection between Hlc and  Our studies also demonstrated a direct correlation between DNA binding affinities and DNA unwinding activities of WRN. The order of DNA binding was HlcRQC Ͼ Hlc-K1016A mutant Ͼ Hlc. This order was the same for DNA unwinding activity. Moreover, unwinding kinetics of HlcRQC closely resemble previously published unwinding kinetics of the full-length WRN using the same 19-bp fork DNA substrate (35). This suggests that the HlcRQC containing aa residues from 500 to 1104 is the minimum requirement for unwinding 19-bp dsDNA.
The database for WRN patient mutation profile indicates that the majority of WS cases is caused by frameshift mutations in the WRN gene as a result of insertion, deletion, nonsense mutation, or splice site variants (48 -50). Although there is no record of WRN patients carrying the Lys-1016 mutation, the importance of Lys-1016 for proper WRN function suggests that such polymorphisms could predispose individuals carrying this mutation to various WS-associated phenotypes and shorten their life span.