Identification and biochemical characterization of a Werner's syndrome protein complex with Ku70/80 and poly(ADP-ribose) polymerase-1.

Werner's syndrome (WS) is an inherited disease characterized by genomic instability and premature aging. The WS gene encodes a protein (WRN) with helicase and exonuclease activities. We have previously reported that WRN interacts with Ku70/80 and this interaction strongly stimulates WRN exonuclease activity. To gain further insight on the function of WRN and its relationship with the Ku heterodimer, we established a cell line expressing tagged WRN(H), a WRN point mutant lacking helicase activity, and used affinity purification, immunoblot analysis and mass spectroscopy to identify WRN-associated proteins. To this end, we identified three proteins that are stably associated with WRN in nuclear extracts. Two of these proteins, Ku70 and Ku80, were identified by immunoblot analysis. The third polypeptide, which was identified by mass spectrometry analysis, is identical to poly(ADP-ribose) polymerase-1(PARP-1), a 113-kDa enzyme that functions as a sensor of DNA damage. Biochemical fractionation studies and immunoprecipitation assays and studies confirmed that endogenous WRN is associated with subpopulations of PARP-1 and Ku70/80 in the cell. Protein interaction assays with purified proteins further indicated that PARP-1 binds directly to WRN and assembles in a complex with WRN and Ku70/80. In the presence of DNA and NAD(+), PARP-1 poly(ADP-ribosyl)ates itself and Ku70/80 but not WRN, and gel-shift assays showed that poly-(ADP-ribosyl)ation of Ku70/80 decreases the DNA-binding affinity of this factor. Significantly, (ADP-ribosyl)ation of Ku70/80 reduces the ability of this factor to stimulate WRN exonuclease, suggesting that covalent modification of Ku70/80 by PARP-1 may play a role in the regulation of the exonucleolytic activity of WRN.

and cancer are the most common causes of death among WS patients. The median age of death is ϳ47 years (1,3). Cells isolated from WS patients show genomic instability and a shorter replicative life span (4). The genomic instability is characterized by an elevated rate of chromosomal translocations and extensive genomic deletions (5). These findings suggest that genomic instability underlies the development of the diseases associated with WS. Cultured cells from WS patients are also hypersensitive to some DNA damaging agents (4), suggestive of a defect in the repair of specific DNA lesions.
Werner's syndrome is caused by mutations within a single gene, which is located on chromosome 8 (6). The cDNA encodes a protein (Werner's syndrome protein, WRN) with strong homology to a class of enzymes called RecQ helicases (7). In addition, the amino-terminal region of WRN is highly homologous to the nuclease domain of Escherichia coli DNA polymerase I and ribonuclease D (8). Helicase and exonuclease activities with a 3Ј to 5Ј directionality have been demonstrated in vitro using recombinant WRN (9 -14). A nuclear localization signal is found near the carboxyl-terminal end of WRN (4). All of the WRN mutations in individuals with Werner's syndrome result in non-sense mutations or frameshifts leading to truncated proteins. The prevailing hypothesis is that the aberrant proteins do not enter the nucleus and are rapidly degraded. Consistent with this idea, cell lines from WS patients show no detectable WRN polypeptide (15).
A number of studies have indicated that WRN binds to proteins that are involved in DNA replication and repair, such as the replication protein A, topoisomerase I, DNA polymerase ␦Fen-1, p53, proliferating cell nuclear antigen, and Rad 52 (11, 16 -22). Although some of these proteins have been shown to influence WRN catalytic activities in vitro, the physiological significance of these interactions remains largely unknown.
In previous studies, we reported that WRN binds to Ku70/80 heterodimer (Ku) (23), a factor involved in the repair of doublestrand DNA breaks by non-homologous end joining (reviewed in Ref. 24). Remarkably, our studies showed that Ku recruits WRN to DNA ends and alters the properties of the WRN exonuclease (23,25). Other studies have also indicated that Ku70/80 is required for the WRN-mediated hydrolysis of DNA molecules containing lesions mimicking oxidative DNA damage (26). A functional interaction between WRN and Ku70/80 is also supported by genetic studies showing that Ku80-null mice display genomic instability and shortened life span (27)(28)(29)(30)(31). Thus, biochemical and genetic evidence suggest that Ku80 and WRN may function together in a DNA repair pathway required for the maintenance of genome integrity.
In this study, we report the identification of a cellular WRN complex composed of WRN, Ku70/80, and poly(ADP-ribose) polymerase-1 (PARP-1). PARP-1 is nuclear factor implicated in the control of genomic stability and mammalian life span (32,33). Our results indicate that a subpopulation of PARP-1 coelutes over ion-exchange and gel-filtration chromatography and coimmunoprecipitates with WRN and Ku70/80. Further biochemical analyses show that PARP-1 poly(ADP-ribosyl)ates Ku70/80 but not WRN in vitro, and ADP-ribosylation of Ku70/80 reduces its DNA-binding activity and weakens its ability to stimulate the exonuclease activity of WRN. To generate pRRLsin.hCMV-Puro, the internal ribosome entry site-green fluorescent protein (IRES-GFP) region of the vector pRRLsin.hCMV-IRES-GFP (35) was replaced by a minimal SV40 promoter driving the expression of the puromycin N-acetyl transferase gene. We produced recombinant lentivirus preparations by three-plasmid transient co-transfection of human 293T cells as described by Naldini et al. (36). For lentiviral infection, 293T cell cultures were trypsinized, seeded onto 60-mm plates, and incubated at 37°C for 24 h. The supernatant containing viral particles was collected and added to cultures that were 30 -40% confluent. After a 6-h incubation at 37°C, the supernatant was removed, the cells were washed twice and incubated in Dulbecco's modified Essential medium containing 10% serum at 37°C. Transduced cells expressing Flag-WRN and Flag-WRN H were selected in media supplemented with puromycin (10 g/ml). The expression of Flag epitope-tagged proteins was analyzed by immunoblotting with anti-Flag antibodies (Sigma).
Purification of the WRN Complex-100 mg of nuclear extracts prepared from cells expressing Flag-WRN H were incubated with anti-Flag beads at 4°C for one hour. After extensive washes, bound proteins were eluted with BCO buffer (1 M KCl, 10 mM Tris HCl, pH 7.5, 1 mM EDTA, 5% glycerol, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 5 g/ml aprotinin) (23), resolved by SDS-polyacrylamide gel electrophoresis, and visualized by silver staining. For the identification of the 120-kDa associated factor by mass spectrometry, we isolated the WRN complex from ϳ1.0 g of nuclear extracts prepared from 293T-WRN H cells. After the immunopurification of the WRN complex, the proteins were separated by SDS-PAGE and stained with Coomassie blue. The gel was extensively destained, and the protein bands were excised from the gel and shipped to the Howard Hughes Medical Institute Biopolymer Facility and W. M. Keck Foundation Biotechnology Resource Laboratory at Yale University for tryptic peptide digestion and sequencing by mass spectrometry. Twenty-four peptides matched PARP-1 amino acid sequences (see Table I).
Immunoprecipitation-Nuclear extracts from 293T cells were prepared as described previously (23). The immunoprecipitation of cellular WRN was carried out using antibody generated by immunization of rabbits with a recombinant WRN amino-terminal fragment (amino acids 1-423). Rabbit immunization and sera production were contracted to an outside company (Bethyl Inc.). Immunoprecipitation products were resolved by SDS-PAGE and transferred to nitrocellulose. Western blot analyses were performed by using antibodies raised against WRN, Ku70, and Ku80 (Santa Cruz Biotechnology), PARP-1 (Santa Cruz Biotechnology) and poly(ADP-ribose) (Trevigen).
(ADP-ribosyl)ation of Proteins-20 g of purified Flag-PARP-1 was incubated with 2 g His-Ku70/80 or His-WRN in a reaction mixture (500 l) containing 100 mM Tris pH 8.0, 10 mM MgCl 2 , 1 mM dithiothreitol, 20 ng/l sonicated salmon sperm DNA, and 20 M NAD ϩ . After incubation at 30°C for 2 h, the salts concentration in the reaction was adjusted to 400 mM KCl, and the reaction mixture was incubated with anti-Flag beads at 4°C for one hour. The supernatant was collected and incubated with metal affinity resin (Talon, Clontech) at 4°C for one hour. After extensive washes with 20 mM Tris, pH 8.0, 150 mM KCl, 1 mM MgCl 2 , 10% glycerol, and a mixture of protease inhibitors, Ku70/80 or WRN were eluted with buffer containing 100 mM imidazole, dialyzed, and analyzed by SDS-polyacrylamide gel electrophoresis, silver staining, and immunoblotting, or snap frozen at Ϫ80°C.

RESULTS
Isolation of WRN-associated Proteins from Cells Stably Expressing Flag-WRN H -To assist in the purification of a native WRN protein complex, we generated stable 293T cells using recombinant, replication-defective, lentivirus vectors expressing either wild-type Flag-WRN or Flag-WRN H , a WRN protein carrying a point mutation (K577M) that inactivates the helicase activity (25,34). Because Flag-WRN is expressed at extremely low levels (Fig. 1A, lane 1), this cell line is not particularly useful for biochemical studies. One possible explanation for this effect is that the increased level of functional WRN may be toxic to the cell. On the other hand, the level of expression of Flag-WRN H (lane 2), which is comparable with that of the endogenous WRN (data not shown), is sufficient for the isolation and biochemical characterization of WRN H from cell extracts. Thus, we prepared nuclear extracts from the 293T-WRN H cells and purified the tagged protein by affinity chromatography on anti-Flag resin. In parallel, extracts from 293T cells that were infected with a control lentivirus were subjected to the same purification procedure. Proteins bound to the affinity column were eluted with high salts and examined by SDS-polyacrylamide gel electrophoresis and silver staining. This analysis revealed the presence of three polypeptides of ϳ70, 90, and 120 kDa, respectively. These proteins were eluted specifically from the resin incubated with the 293T-WRN H nuclear extract and were absent from the eluate of the control resin (Fig. 1). The 70-and 90-kDa polypeptides were identified as Ku70 and Ku80, respectively, by immunoblot analysis (Fig.  1C). To identify the 120-kDa polypeptide, the protein band was excised from the SDS-polyacrylamide gel, subjected to proteolytic digestion, and analyzed by matrix assisted laser desorption ionization mass spectrometry. Data base searches indicated that the 120-kDa polypeptide is identical to PARP-1 (see Table I), an enzyme that is activated by DNA damage and utilizes NAD ϩ to catalyze the addition of poly(ADP-ribose) on target proteins. The initial identification was confirmed by immunoblot analysis with monoclonal anti-PARP-1 antibody (Fig. 1C).
Physical Interaction between WRN and PARP-1 in Human Cells-To provide further evidence that PARP-1 interacts with endogenous WRN in vivo, we immunoprecipitated WRN from nuclear extracts prepared from 293T cells using anti-WRN antibodies. As a control, the same nuclear extract was subjected to immunoprecipitation with antibody against ␤-actin. The products of the immunoprecipitation reactions were resolved by SDS-polyacrylamide gel electrophoresis and analyzed by immunoblotting. The results indicate that PARP-1 coimmunoprecipitates with WRN but not with ␤-actin ( Fig. 2A). The Ku70/80 complex is also detected in the immunoprecipitation reaction with anti-WRN antibodies, confirming the results of the immunopurification of Flag-WRN H . Treatment of the nuclear extract with DNase I prior to the immunoprecipitation yielded identical results (data not shown).
Previous studies have shown that PARP-1 binds to Ku70/80 (38, 39); therefore, it is possible that Ku70/80 mediates the interaction between WRN and PARP-1. To determine whether WRN binds directly to PARP-1, recombinant Flag-PARP-1 was immobilized on anti-Flag beads and incubated with recombinant WRN (His-WRN). In parallel reactions, immobilized Flag-PARP1 was incubated with Ku70/80 or with a mixture containing Ku70/80 and WRN. After washing the beads, the bound proteins were resolved by SDS-PAGE and analyzed by Western blotting with antibodies against WRN, Ku70/80, and PARP-1. As shown in Fig. 2B, WRN binds to PARP-1 in the absence of Ku70/80 (lane 6), suggesting that the interaction between WRN and PARP-1 in the WRN complex is mediated, at least in part, by direct interaction between these two proteins. Similarly, WRN binds to PARP-1 in the presence of Ku70/80 (lane 8). These results suggest that these factors can individually associate with each other and can form a trimeric complex in vitro.
Analysis of the WRN Complex by Ion Exchange and Gel Filtration Chromatography-To further examine the relationship among WRN, Ku70/80, and PARP-1 and determine whether there is evidence for the existence of a stably associated complex in vivo, we performed a three-step fractionation of the nuclear extracts using a DEAE ion-exchange column, SP-Sepharose ion-exchange column, and Superose 6 gel-filtration columns (Fig. 3). The flow-through of the DEAE column, which contains PARP-1, WRN, and Ku70/80 as detected by immunoblotting (Fig. 3B, lane 2), was applied to an SP-Sepharose column. Immunoblot analysis of the fractions from the step-  1 and 2, respectively) (lanes 1 and 3) or buffer only (lanes 2 and 4). Ku70/80 were collected by incubation with a metal ion resin, solubilized in SDS-containing buffer, resolved by SDS-PAGE, and analyzed by silver staining (lanes 1 and 2) and analysis also indicates that a minor fraction of Ku70 elutes in the void volume, possibly suggesting the presence of high molecular weight Ku complexes or aggregates.
A similar analysis of the fractions eluted from the Superose 6 column loaded with the SP-Sepharose 0.4 M KCl fraction (SP-0.4 Superose 6) shows that WRN coelutes with a portion of PARP-1 and Ku in fractions 44 -52 (Fig. 3C, bottom panel). Note that these data are only suggestive of complex formation. Complex formation per se was then tested by direct immunoprecipitation of these fractions by using anti-WRN antibodies and by probing the precipitated proteins for immunoreactivity against WRN, PARP-1, and Ku70. As shown in Fig. 3D, Ku70 coimmunoprecipitated with WRN from pooled fractions 45/46 and fractions 47/48 of the SP-0.15 Superose 6 column (Fig. 3D,  lanes 1 and 2), and Ku70 and PARP-1 coimmunoprecipitated with WRN from pooled fractions 45/46 and fractions 47/48 of the SP-0.4 Superose 6 column (lanes 3 and 4). These results suggest that two WRN complexes may exist in vivo, one with Ku and another with both Ku and PARP-1.
Poly(ADP-ribosyl)ation Reduces Ku70/80 DNA-binding Activity-PARP-1 is a nuclear protein that binds to DNA strand breaks and catalyzes ADP-ribosylation of itself and other nuclear proteins by using NAD ϩ as a cofactor. This catalytic activity requires DNA. To establish that our preparation of purified recombinant PARP-1 was able to poly(ADP-ribosyl)ate itself in a DNA-dependent manner, we incubated recombinant PARP-1 in the absence or presence of increasing amounts of DNA and NAD ϩ . As shown in Fig. 4A, recombinant PARP-1 becomes poly(ADP-ribosyl)ated in the presence but not in the absence of DNA, as indicated by the appearance of high molecular mass products in the silver-stained gel (compare Fig. 4A,  lanes 1 and 2 with lane 3). Importantly, poly(ADP-ribosyl)ation of PARP-1 depends on the presence of NAD ϩ , as incremental formation of high molecular mass products is observed in the presence of increasing amounts of NAD ϩ , and the omission of NAD ϩ from the reaction mixture prevents the formation of these products (Fig. 4A, lanes 4 -7). To determine whether WRN and Ku70/80 were substrates of the poly(ADP-ribose) polymerase activity of PARP-1, purified WRN and Ku70/80 were individually incubated with PARP-1 in the presence of NAD ϩ and DNA. WRN and Ku70/80 were then immunoprecipitated from the respective reaction mixtures, analyzed by SDSpolyacrylamide gel electrophoresis, and immunoblotted with a monoclonal antibody that recognizes poly(ADP-ribosylation) (anti-PAR antibody). The results of this experiment indicate that PARP-1 poly(ADP-ribosyl)ates Ku70/80 (Fig. 4B, lane 8) but not WRN (Fig. 4B, lane 6). Interestingly, the apparent molecular mass of Ku70 and Ku80 does not change significantly after covalent modification, suggesting that only a few ADP-ribose molecules are added to both Ku70 and Ku80.
Poly(ADP-ribosyl)ation Inhibits Ku70/80 DNA-binding Activity and Reduces the Ku-dependent Stimulation of WRN Exonuclease Activity-The addition of ADP-ribose molecules increases the negative charge of a protein, which is thought to alter the DNA-binding properties of the target factor. To deter-mine whether ADP-ribosylation influences the DNA-binding activity of Ku70/80, we purified in vitro poly(ADP-ribosyl)ated Ku70/80 and unmodified Ku70/80. Ku70/80 was incubated with PARP-1 in the presence of DNA and NAD ϩ to allow poly(ADPribosyl)ation and purified by affinity chromatography. The purity of unmodified and poly(ADP-ribosyl)ated Ku70/80 was determined by silver staining of SDS-polyacrylamide gels (Fig.  5A, lanes 1 and 2), and poly(ADP-ribosyl)ation was confirmed by immunoblot analysis with anti-PAR antibody (lanes 3 and  4). Increasing amounts of unmodified or poly(ADP-ribosyl)ated Ku70/80 (Fig. 5A) were incubated with a radiolabeled doublestrand oligomer, and DNA-binding activity was analyzed by electrophoretic mobility shift assays (Fig. 5, B and C). The results of this experiment indicate that ADP-ribosylation significantly reduced the DNA-binding activity of Ku70/80 (compare lanes 2-4 to 5-7), suggesting that this covalent modification alters the DNA-binding properties of this factor.
We have shown previously that Ku70/80 recruits WRN to DNA ends and stimulates WRN exonuclease activity. Because PARP-1 has been reported to bind to DNA ends (40), we examined whether PARP-1 may influence WRN exonuclease activity. We performed WRN exonuclease assays with PARP-1 in the presence and absence of Ku70/80 and found that unmodified PARP-1 or poly(ADP-ribosyl)ated PARP-1 do not affect the exonuclease activity of WRN nor alter the strong stimulation of this activity by Ku70/80 (data not show). Then we wanted to determine whether poly(ADP-ribosyl)ation of Ku70/80 influences WRN exonuclease activity. Because the experimental conditions used for the exonuclease assay are not suitable for PARP-1 poly(ADP-ribosyl)ation activity, Ku70/80 was poly-(ADP-ribosyl)ated in vitro prior to its addition to the exonuclease reaction (Fig. 5A). WRN was then incubated with a radiolabeled double-strand oligonucleotide in exonuclease buffer in the presence of poly(ADP-ribosyl)ated Ku70/80 or unmodified Ku70/80. The products of these reactions were separated by denaturing acrylamide gel electrophoresis and analyzed by autoradiography and PhosphorImager analyzer. The results of these experiments show that poly(ADP-ribosyl)ated Ku70/80 leads to a modest increase in DNA hydrolysis as compared with the unmodified factor (Fig. 5, panels D and E), indicating that poly(ADP-ribosyl)ation reduces the ability of Ku70/80 to stimulate the exonuclease activity of WRN. DISCUSSION In this study, we purified a WRN complex by affinity chromatography and identified its components by immunoblot analysis and mass spectroscopy. We found that WRN resides in a complex with Ku70/80 and PARP-1. Coimmunoprecipitation assays indicated that PARP-1 binds to WRN and Ku70/80 in vivo, and in vitro protein-binding studies with purified factors show that the interaction between PARP-1 and WRN is direct and that WRN can form a trimeric complex with PARP-1 and fication of Ku70/80 as one of the components of the WRN complex is consistent with previous studies (23,25,43), which established Ku70/80 as a functional partner of WRN and suggested a possible role for WRN in DNA repair. The finding that PARP-1, a factor implicated in the cellular response to DNA damage, is a component of the WRN complex supports the idea that WRN may be involved in a pathway that monitors genome integrity. This interpretation is in agreement with a recent genetic study, which showed that double knockout mice lacking WRN and PARP-1 display chromosomal instability and shorter life span (44).
PARP-1 is a highly conserved eukaryotic protein that binds to single-and double-strand DNA breaks and is thought to function as a sensor of DNA damage (32,33). In response to DNA damage induced by ionizing radiation, alkylating agents, and oxidants, PARP-1 binds to DNA. DNA binding causes the activation of PARP-1 catalytic activity, which utilizes NAD ϩ to catalyze the addition of multiple ADP-ribose molecules to acceptor proteins. A number of studies have reported that activated PARP-1 poly(ADP-ribosyl)ates proteins such as histones, p53, topoisomerases, lamins, and PARP-1 itself (45)(46)(47)(48)(49)(50). Poly-(ADP-ribosyl)ation of p53 has been reported to influence the DNA-binding activity of this tumor-suppressor protein (51). Moreover, poly(ADP-ribosyl)ation of chromosomal proteins has been proposed to alter the nucleosomal structure near the DNA strand breaks and to promote the access of repair enzymes to these sites (52). We have examined whether WRN and Ku70/80 are substrates of PARP-1 enzymatic activity and have shown that PARP-1 poly(ADP-ribosyl)ates Ku70 and Ku80 but not WRN. In addition, our analysis indicates that poly(ADP-ribosyl)ation of Ku70/80 alters the DNA-binding activity of this factor and inhibits the Ku70/80-mediated stimulation of WRN exonuclease activity. Thus, these results indicate that, through covalent modification of Ku70/80, PARP-1 modulates WRN exonuclease activity. Conversely, we did not observe any significant alteration of WRN helicase activity by PARP-1 in the presence or absence of Ku70/80 (data not shown). Given the role of Ku70/80 and PARP-1 in the recognition of DNA breaks, our findings suggest that WRN, once recruited to the site of DNA damage, may participate directly in the processing and resolution of the broken DNA ends. Because there are severalfold more Ku and PARP-1 molecules than WRN in a cell, it is conceivable that WRN in the context of this complex functions as a nucleating factor that organizes a subset of Ku70/80 and PARP-1 molecules into a DNA damage sensory complex involved in the surveillance of genome integrity.
Our fractionation studies indicate that WRN co-elutes with subpopulations of Ku70/80 and PARP-1. It is important to note that coincidence of the peaks of each factor in the gel-filtration chromatography is not expected, as this would suggest that the majority of all three proteins exist in a complex. Such a result is not anticipated, as it is known that all three proteins interact with different partners in response to different physiological cues (24,53,54). Importantly, to verify that complex formation was not an artifact of the three factors independently binding to DNA, immunoprecipitation experiments were carried out in the presence of DNase I. These experiments yielded identical results (data not shown). The immunoprecipitation of WRN from eluates of fractionated extracts show that WRN forms two complexes with Ku, one of which contains PARP-1. These findings may indicate that there are two distinct WRN-Ku complexes in the cell. Alternatively, the association of PARP-1 with Ku70/80 and WRN may be part of a dynamic process that involves the rapid assembly and disassembly of this complex in the cell in response to diverse physiological and pathological stimuli. Clearly, we cannot rule out that the two observed WRN-Ku complexes may only be a consequence of the fractionation experiments; future studies will address this issue.
PARP-1 has been implicated in the protection of genome integrity by facilitating the repair of damaged DNA. In addition, PARP-1 participates in the execution of the apoptotic and necrotic pathways in cells with excessive DNA damage (55)(56)(57). The necrotic function of PARP-1 has been linked to overactivation of its enzymatic activity, which leads to the cellular depletion of NAD ϩ and ATP (32,54,55). Excessive PARP-1 activity has been implicated in the pathogenesis of several clinical conditions such as stroke, myocardial infarction, arthritis, diabetes, and neurodegenerative disorders (54). The molecular mechanisms that regulate PARP-1 survival and death-promoting effects are poorly defined; thus, it is possible that WRN and Ku70/80 may play a role in the selection of the specific pathway that is activated in response to DNA damage.
PARP-1 and poly(ADP-ribosyl)ation have been linked to mammalian longevity, as differences in the catalytic activity of PARP-1 correlate with differences in life span between short and long-lived species (58). Because WS is a premature aging disease, and inactivation of Ku in mice leads to premature aging (27,28,31), the identification of a physical interaction between WRN, Ku70/80, and PARP-1 suggests that these proteins are caretakers that function together in a cellular pathway that monitors the integrity of the genome and the longevity potential of an organism. Future biochemical and genetic studies will help in elucidating the link between the WRN complex and the network of cellular pathways that control the fate of cells with damaged DNA and shortened life spans.