| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 278, Issue 38, 36476-36486, September 19, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
**
From the
Centre for Molecular Biology and Neuroscience, and Institute of Medical Microbiology, University of Oslo, Rikshospitalet, 0027 Oslo, Norway, the ¶Institute of Cancer Research and Molecular Biology, Norwegian University of Science and Technology, Olav Kyrresgate 3, N-7005 Trondheim, Norway, and the ||Laboratory of Molecular Gerontology, NIA, National Institutes of Health, Baltimore, Maryland 21224
Received for publication, April 14, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Cells from Werner syndrome (WS) patients exhibit elevated genome instability manifested as telomere defects, and chromosomal translocations and large deletions (6, 7, 8). The latter rearrangements are thought to reflect aberrant DNA recombination, a process responsible for the repair of single strand gaps and double strand breaks (DSBs). Such DNA damage can arise directly from exposure to ionizing radiation or some chemicals and indirectly as a consequence of replication fork blockage (see Ref. 9 and references therein). Two major recombination pathways have been identified in eukaryotes that are distinct with respect to mechanism and DNA homology requirements. The non-homologous end-joining (NHEJ) pathway joins two extremities of a DSB via a process that is largely independent of terminal DNA sequence homology. Therefore, NHEJ can be error-prone, producing deletions, insertions, and translocations (10, 11). Homologous recombination (HR) corrects strand breaks using homologous sequences primarily from the sister chromatid and, to a lesser extent, from the chromosome homologue (12, 13); therefore, it is a high fidelity repair mechanism.
HR is poorly defined in eukaryotes, and the biological roles of specific proteins are unclear. However, several lines of evidence indicate that the principle steps of HR are likely conserved from yeast to humans. In S. cerevisiae, HR is carried out by proteins encoded by the RAD52 epistasis group, which includes the key Rad51p recombinase and the Rad52p recombination mediator/single strand annealing protein (10, 11, 14). Like the yeast protein, human RAD52 binds as heptameric rings to single strand regions and DSBs (15, 16), an action that may facilitate RAD51 loading at the recombination site (17). Once bound to the DNA, RAD51 forms a nucleoprotein filament that catalyzes strand exchange between a 3' terminus and a donor duplex. The exchange reaction is stimulated by RAD52 and produces a joint molecule or Holliday junction. RAD52 also promotes the homologous strand pairing and exchange reaction independently of RAD51 (14). Despite strong similarities between amino acid sequence and biochemical activities pointing to a conservation of biological function, the cellular roles of the yeast and human RAD51 and RAD52 homologous proteins are to some extent divergent. Rad52p-defective yeast cells exhibit a severe and general recombination defect, whereas the lack of vertebrate RAD52 only partially reduces the frequency of HR events (see Ref. 18 and references therein), leading to speculation about the presence of a backup or redundant activity. Although the absence of Rad51p is hardly noticeable in yeast, a RAD51 deficiency in vertebrates produces an embryonic lethal phenotype (19, 20, 21).
The chromosomal aberrations exhibited by WS cells resemble those observed in vertebrate cells defective for HR proteins (22–24). Indeed, recent evidence has correlated WRN-deficiency with specific HR defects. WS cell lines have been shown to be proficient at initiating mitotic recombination but defective at resolving recombination intermediates (22). A potential substrate for WRN was revealed when expression of RusA, a bacterial protein cleaving almost exclusively four-way junctions in vitro, rescued cell survival and restored the ability to generate viable recombinants following exposure of WS to DNA-damaging agents (8). These results point to the direct participation of WRN in HR. The yeast RecQ homologue, Sgs1p, has also been associated with HR, pointing to an evolutionary conservation of function. Similar to WS cells, Sgs1p-deficient strains exhibit erroneous homologous recombination (25). Sgs1p demonstrates epistasis with Rad51p and Rad52p and plays an important role in modulating Rad52p-dependent spontaneous and induced recombination events (26, 27). Furthermore, Sgs1p has been shown to physically associate with Rad51p (28). Finally, recent work has shown that Sgs1p, together with the Mus81/MMS4 endonuclease and the Srs2 helicase, function on alternate pathways, in coordination with recombination protein activities, to prevent the formation of toxic recombination intermediates at stalled replication forks (29).
In the present study, we wanted to further investigate the reported role of WRN in HR by examining the possibility of there being a direct association between WRN and the human counterparts of the key yeast HR proteins, RAD51 and RAD52. We report that WRN interacts directly with RAD52 both in vitro and in vivo. Fluorescence resonance energy transfer (FRET) analyses provide additional support for this interaction by showing that fluorescently tagged RAD52 and WRN proteins are in a complex resembling blocked replication forks in response to DNA damage in vivo. Biochemical analyses demonstrate that RAD52 modulates WRN helicase activity in a manner that appears to depend on the substrate configuration. WRN, in turn, stimulates RAD52-mediated homologous strand annealing between complementary sequences in a double-stranded plasmid DNA and single-stranded oligonucleotide. These results shed further light on the poorly understood processes of replication fork recovery in human cells by suggesting that WRN functions together with RAD52 in response to replication forks attenuated by DNA damage.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Plasmids—The pBlueBac-WRN baculovirus used for overexpression of the full-length WRN protein in Sf9 cells has been previously described (30). DNA fragments encoding the WRN C terminus (WRN-C; amino acids 982–1432) and exonuclease truncated peptide (WRN-EXO) were PCR-amplified from the pBlueBac-WRN template and cloned into the EcoRI-BamHI and BglII sites, respectively, using standard cloning techniques (31). The RAD52 and RAD51 DNA sequences were PCR-amplified for cloning into the BamHI site of the yeast two-hybrid vectors pGAD424 and pGBT9 (MATCHMAKER two-hybrid system, Clontech Laboratories) from pFB581 and pFB530 (32, 33) (generous gifts from F. Benson, Imperial Cancer Research Fund, South Mimms, UK). All clones were verified to be in-frame by sequencing.
Yeast Two-hybrid Assays—The yeast strain used in this study, PJ69-4A, has been described previously (34) and was a kind gift from O. Gabrielsen (University of Oslo, Norway). The strain carries the HIS3 and ADE2 reporter genes, enabling growth assays on selective media with different stringencies. Transformation with the respective combinations of GAL4 activation (AD: pGAD424 and LEU2+) and DNA binding domain constructs (BD: pGBT9 and TRP+) was performed using an Sc EasyComp yeast transformation kit (Invitrogen) according to the manufacturer's instructions. Transformants were plated onto yeast nitrogen base ((Difco) 2% glucose, 2% gelose) medium lacking Leu/Trp (YNB-Leu/Trp, Clontech) and incubated for 2 days at 30 °C. Single transformants were re-suspended in sterile water at equal densities, spotted onto YNB-Leu/Trp (control), YNB-His/Leu/Trp/+3-amino-1,2,4,-triazole (1.5 mM) (Sigma Chemical Co.) (weaker interactions), and YNB-Ade/His/Leu/Trp (stronger interactions) and scored for differential growth after 3–5 days of incubation at 30 °C.
Antibodies—Polyclonal rabbit anti-RAD52 (H-300) and goat anti-RAD52 (C-17) antibodies, rabbit IgG, and pre-immune serum were purchased from Santa Cruz Biotechnology Inc. The anti-WRN (ab200) antibody was obtained from Abcam (Cambridge, UK). Anti-rabbit and anti-goat alkaline phosphatase-conjugated secondary antibodies (H+L) were purchased from Roche Molecular Biochemicals.
Protein Expression and Purification—Purification of the full-length, hexahistidine-tagged WRN protein from Sf9 cultures has been described previously (30). BL21-CodonPlus(DE3)-RIL cells transformed with the hexahistidine-tagged RAD52 expression construct pFB581 were grown in LB (0.1 mg/ml ampicillin), harvested following a 2-h induction with 1 mM isopropyl-
-D-thiogalactopyranoside (Fermentas Inc.) as recommended (Qiagen) and sonicated in a lysis buffer (50 mM Tris, pH 8, 1 M NaCl, 5 mM -mercaptoethanol). The resulting lysate was loaded directly onto a nickel-nitrilotriacetic acid column (Qiagen) and RAD52 was eluted with lysis buffer containing 0.3 M imidazole. Overexpression and purification of the N-terminally truncated version 
N1–400 of the S. cerevisiae RecQ homologue Sgs1 from vector pRB222 (a generous gift from J.C. Wang) was performed as described previously (35). All protein preparations were tested for purity and activity prior to use.
Far Western Analysis—Typically, 0.5–1 µg of active, recombinant protein was subjected to SDS-PAGE (8%) and transferred to Immobilon-P polyvinylidene difluoride membranes (Millipore). All steps were performed at 4 °C unless otherwise indicated. Filters were incubated overnight in a buffer comprised of 10 mM Tris, pH 7.4, 0.5% BSA, 0.25% gelatin, 0.2% Triton X-100, 5 mM -mercaptoethanol, and 150 mM NaCl. Fresh buffer was added with or without 1 µg of the respective partner protein, and the membrane was incubated for a further 4–6 h, at which time it was washed briefly at room temperature with buffer containing 10 mM Tris-HCl, pH 8, 0.05% Tween, and 150 mM NaCl. Conventional Western blot analysis was performed to detect the presence of WRN or RAD52 using WRN- and RAD52-specific antibodies.
Immunoprecipitation—All steps were performed at 4 °C unless otherwise stated. Potential interactions between recombinant, full-length WRN and RAD52 were investigated by incubating 1 µg of each protein together for 1 h in immunoprecipitation buffer (50 mM HEPES, pH 7.5, 5 mM MgCl2, 150 mM NaCl, 0.05% Triton X-100, 1 mM EDTA, 1 mM PMSF, and 1 mM DTT), followed by a 1-h incubation with either anti-WRN antibody, anti-RAD52 antibody (H-300), or rabbit IgG, together with Protein G-Sepharose Fast Flow 4 beads (Amersham Biosciences). Immunocomplexes were pelleted, washed four to five times with immunoprecipitation buffer, re-suspended in protein loading buffer (2% SDS, 5% -mercaptoethanol, 20% glycerol, 60 mM Tris, pH 6.8, and 0.01% bromphenol blue), and subjected to conventional Western blotting analysis. The RAD52 blots for were probed with the C-17 antibody to minimize cross-reactivity between the rabbit IgG heavy chain and the goat IgG heavy chain, as the rabbit heavy chain runs close to the 50-kDa protein marker rendering separation from RAD52 difficult.
Nuclear extracts for co-immunoprecipitation experiments of endogenous proteins were prepared in the following manner from asynchronously growing HeLaS3 cultures. All steps were performed on ice unless otherwise indicated. Approximately 80% confluent cultures (1 x 107 cells) were washed twice with phosphate-buffered saline, and the cells were collected with a cell scraper and re-suspended in 2 ml of phosphate-buffered saline. Cells were pelleted, re-suspended in two volumes of lysis buffer A (10 mM HEPES, pH 7.9, 0.5 mM MgCl2, 10 mM KCl, 1 mM DTT, 0.2 mM PMSF) and incubated for 15 min on ice. Whole cell extracts were centrifuged in buffer A suspension for 15 min at 2,500 x g, 4 °C, and the resulting pellets were re-suspended in 2 volumes of buffer B (20 mM HEPES, pH 7.9, 0.5 mM MgCl2, 0.42 mM NaCl2, 25% glycerol, 0.2 mM EDTA, 1 mM DTT, 0.2 mM PMSF). A final centrifugation step for 30 min at 14,000 x g, 4 °C, yielded the nuclear extract in the supernatant. Total protein in the final nuclear extract was determined by conventional Bradford protein assay (Bio-Rad) and was typically 150–300 µg.
Immunoprecipitation of protein complexes from HeLaS3 nuclear extracts was carried out by diluting samples to 200 µl with dilution buffer containing 50 mM Tris-HCl, pH 8, 150 mM NaCl, 0.5% Nonidet P-40, 1 mM DTT, 1 mM PMSF. Following preclearing with protein G-Sepharose, extracts were incubated 2 h at 4 °C with 0.5 µg each of anti-RAD52 or anti-WRN antibodies or 0.5 µg of pre-immune rabbit serum (Santa Cruz Biotechnologies). Immunocomplexes were captured by incubating 1/10 of the total volume of protein G-Sepharose with the extracts for 2 h at 4 °C. After extensive washing of the protein G-Sepharose with dilution buffer, the pellets were re-suspended in protein loading buffer, separated by 8% SDS-PAGE, and subjected to conventional Western blotting analysis. The H-300 antibody was used to evaluate RAD52 immunoprecipitation and co-precipitation in these experiments.
Cloning of Fusion Proteins—EYFP-WRN was made from EGFPC3-WRN (35) by cloning the AgeI-SalI fragment from the pEYFP-C1 vector (Clontech) into the AgeI-XhoI sites of EGFPC3-WRN by standard techniques (31). RAD52 was PCR-amplified from pFB581 and ligated into the BamHI restriction site of pECFP-C1 vector (Clontech). The pGFPCNAL2 construct (37) (a generous gift from H. Leonhardt and C. M. Cardoso, Berlin, Germany) was used to prepare the expression constructs pECFP-PCNA (Clontech) by PCR-amplifying and cloning the SV40 nuclear localization signal into the NheI-AgeI sites, followed by ligating the PCNA-containing BsrGI-XbaI fragment from pGFPCNAL2 into the pECFP-C1 vector containing the SV40 nuclear localization signal. pUNG2-ECFP and pUNG2-EYFP constructs were prepared by cloning the AgeI-NotI fragment from pECFP-N1 and pEYFP-NI (Clontech) into the pUNG2EGFP construct (37). All constructs were verified to be in-frame by sequencing.
Confocal Microscopy and FRET Measurements—A Zeiss LSM 510 laser scanning microscope equipped with a Plan-Apochromate 63x/1.4 oil immersion objective was used to examine images of 1-µm-thick slices of living cells. Fluorescence energy transfer (FRET) was determined by modifying the general equations given by Matyus (39) for use with the ECFP and EYFP fusion proteins as donor (D) and acceptor (A), respectively. FRET occurs if I2 – I1[ID2/ID1] – I3[IA2/IA3] > 0, where I represents intensities in three channels given in arbitrary units (see also Ref. 40). Intensities were measured as follows: channel 1: I1, A1, D1 = excitation (ex.) at (
= 458, detection (det.) at 475 nm < (>525 nm (ECFP); channel 2: I2, D2, A2 = ex. at ( = 458 nm, det. at (>560 nm; channel 3: I3, D3, A3 = ex. at ( = 514 nm, det. at (>560 nm (EYFP). ID1, D2, D3 and IA1, A2, A3 are determined separately for cells transfected with only ECFP and EYFP fusion proteins, respectively, under the same settings and at the same levels of fluorescence intensities (I1 and I3) as cotransfected cells. FRET values were normalized to account for differences in the respective fluorochrome expression levels using the following equation: Normalized FRET (NFRET) = FRET/(I1 x I3)1/2 (41).
Substrate Preparation and Biochemical Analyses—The structures of the DNA substrates used for biochemical analysis and the oligonucleotides used to prepare them are presented below in Tables II and III, respectively. Substrates were prepared as described in Mohaghegh et al. (42).
|
|
Labeled oligonucleotides for strand annealing assays (oligonucleotide 4, Table III; substrate 8, Table II) were fractionated on a 20% denaturing PAGE-taurin gel containing 8.3 M urea (2.5 h, 1800 V) and subsequently isolated as gel slices, which were eluted overnight at 37 °C into a buffer containing 0.5 M sodium acetate and 1 mM EDTA. The single-stranded DNA was recovered in the eluate and precipitated with 3 V of a 100% ethanol/0.1 M sodium acetate solution. Pellets were re-suspended in sterile water and were subjected to scintillation counting for final concentration determination.
All biochemical assays were performed using a minimum of two independent protein preparations at the indicated amounts and 1 nM substrate (unless otherwise indicated). The gel images show typical results evaluated using PhosphorImager/ImageQuaNT software (Amersham Biosciences). Exonuclease reactions were performed as previously described (43) with the exonuclease substrate (substrate 1, Table II). One unit of E. coli DNA polymerase I Klenow fragment (New England BioLabs) was used as a control for exonuclease assays. Helicase assays were performed essentially as described in Fry et al. (44) except that the reactions were performed for 30 min at 37 °C. Where indicated, ATP was excluded or substituted with 2.5 mM ATP
S (Roche Applied Science).
WRN modulation of RAD52-mediated strand D-loop formation was performed essentially as described (45) with the following modifications. Specified amounts of RAD52 or RecA (New England BioLabs) were incubated at 37 °C with 1 nM of 5'-radioactively labeled oligonucleotide (Table II) for 10 min in the helicase reaction buffer with or without ATP as indicated. 1 nM of supercoiled pUC19 plasmid was subsequently added, and the reaction was carried out for a further 15 min. WRN at the indicated amounts was subsequently added, and the reactions were incubated for another 20 min. Reactions were terminated by adding final concentrations of SDS and proteinase K of 0.67% and 1 mg/ml, respectively, and the reaction products were analyzed on 1% agarose gels (0.5x TBE running buffer; 70 V, 50 min), which were incubated in a 0.5x TBE solution containing 50 µg/ml ethidium bromide prior to drying to visualize the input DNA as a loading control.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
WRN Associates with RAD52 in Vitro—To verify the two-hybrid results, far Western blot analyses were performed using purified WRN and RAD52 proteins. WRN migrated as a 160-kDa band (Fig. 2A, lane 2), whereas RAD52 primarily self-associates into complexes that are >250 kDa (Fig. 2A, lane 3) (16, 47). Heat treatment prior to sample loading resolves these complexes into monomers running close to 50 kDa (Fig. 2A, lane 4) (47). WRN clearly associates with both monomeric and multimeric forms of immobilized RAD52 (Fig. 2B, lane 3), and conversely, soluble RAD52 binds to immobilized WRN (Fig. 2C, lane 3). These interactions were specific, because neither WRN nor RAD52 bound to immobilized BSA or to purified DNA glycosylases, hNTH1 (human endonuclease III) and hOGG1 (8-oxo-guanine DNA glycosylase 1) involved in the initial steps of the base excision repair of oxidative damage (data not shown). Furthermore, neither immobilized RAD52 nor WRN incubated with interaction buffer alone cross-reacted with the anti-WRN or anti-RAD52 antibodies, respectively (Fig. 2, B and C, respectively, lane 2).
|
Co-immunoprecipitation experiments using the purified proteins were performed to confirm the interaction by an independent method. WRN co-precipitated with immunoprecipitated RAD52 and vice versa (Fig. 2, D and E, lane 3). Coprecipitation depended on the presence of the partner as neither protein was precipitated with rabbit immunoglobulin (Fig. 2, D and E, respectively, lane 2). Furthermore, WRN and RAD52 could not be precipitated with the anti-RAD52 and anti-WRN antibodies, respectively (data not shown). Thus, WRN and RAD52 specifically interact with each other in vitro.
WRN and RAD52 Co-precipitate from Nuclear Extracts—To investigate whether endogenous WRN and RAD52 associate, co-immunoprecipitation experiments were performed using HeLa nuclear extracts. As anticipated, the anti-WRN and -RAD52 polyclonal antibodies efficiently immunoprecipitated their respective, cognate protein (Fig. 3, A and B; lane 3). WRN was co-precipitated with immunoprecipitated RAD52 (Fig. 3A; lane 4). In the corollary experiment, RAD52 precipitated when anti-WRN antibodies were added to the extracts, although blots were difficult to evaluate due to problems encountered in separating endogenous RAD52 (48 kDa) from the rabbit IgG heavy chain (IgGHC) (50 kDa) (Fig. 3B). Similar findings were obtained using nuclear extracts prepared from SK-HEP-1 cells (data not shown). These results show that endogenous WRN and RAD52 are in the same complex in vivo.
|
WRN and RAD52 Co-localize and Exhibit FRET in Vivo— Co-expression of RAD52 and full-length WRN as C-terminal fusions to variants of the enhanced green fluorescent protein (ECFP and EYFP, respectively) allowed co-localization and fluorescence resonance energy transfer (FRET) determinations in living, cycling cells. As reported previously (36), fluorescently tagged WRN localized to nuclei of untreated HeLa cells and accumulated in both nucleoli and nucleoplasmic foci of varying sizes (Fig. 4A). Co-transfection of WRN with fluorescently tagged PCNA, a protein known to accumulate in replication foci (37), showed that most of the nucleoplasmic spots were not replication sites in untreated cells (Fig. 4A, upper row). The RAD52 fusion protein exhibited primarily nucleoplasmic staining but, unlike WRN, did not accumulate in nucleoli (Fig. 4A, lower row). Interestingly, co-transfected RAD52 and WRN co-localized to nucleoplasmic foci in untreated cells (Fig. 4A, lower row), suggesting that both proteins function at the same site in some aspect of DNA metabolism in the absence of exogenous DNA-damaging agents.
|
Several studies have reported increased WRN and RAD52 foci formation following exposure to DNA-damaging agents (48–50). In HeLa cells treated with the intra-strand cross-linking agent mitomycin C (MMC), which blocks replication in vitro (51), no increase in WRN foci was observed after 4–5 h (data not shown). However, following 16–18 h of MMC treatment, a substantial proportion of the cells had WRN foci that co-localized with PCNA, indicating that they represented arrested replication forks (Fig. 4B, upper row). Under identical experimental conditions, RAD52 formed foci that co-localized with the WRN foci (Fig. 4B, lower row), indicating that RAD52 and WRN co-localize to replication forks in response to DNA damage.
FRET analysis was performed to determine if fluorescenttagged RAD52 and WRN proteins interact in vivo or are close within the same protein complex (i.e. <100 Å) at the arrested replication forks (Fig. 4C). FRET occurs only when the intensity of emitted light measured in the presence of two fluorescently tagged proteins is greater than that emitted from cells transfected with only the blue- or yellow-tagged proteins alone (i.e. background levels). Fig. 4C shows intensities along a line through two replication foci (arrows). Table I gives FRET and normalized FRET (NFRET) values (the highest five and one medium low) calculated from the mean of intensities within a replication focus (the region of interest) that were found within the given levels of intensities. Varying FRET levels among different replication foci are to be expected because of the dynamic nature of these interactions in living cells. Equal, or higher FRET values were detected between RAD52 and WRN compared with WRN and PCNA (Table I), which have been reported to interact in vitro (52). FRET was also determined for ECFP- and EYFP-tagged UNG2, a protein localizing to replication foci (53), where it binds as a monomer to PCNA, as a negative control for a direct interaction but positive for colocalization. Values were smaller, but positive for a small fraction of replication foci and thought to reflect UNG2 binding adjacently to trimeric PCNA (37) indicating that the differentially tagged proteins may be within the same protein complex without directly binding to each other. However, in the majority of these foci FRET was not detected (Table I). NFRET for WRN and RAD52 were equal to or higher than values determined for WRN and PCNA, and both sets of NFRET values were 3- to 4-fold higher than those determined for the UNG2 monomers (Table I), thus indicating a direct interaction between WRN and RAD52, as well as WRN and PCNA at arrested replication forks in vivo. Essentially identical results were obtained using the hamster cell line CHO-KI following methyl methanesulfonate treatment for 16–18 h (data not shown). These results further indicate that the interaction between RAD52 and WRN is neither cell type- nor DNA-damaging agent-specific but, rather, is a general response to replication forks arrested by DNA damage.
|
RAD52 Modulates WRN Helicase in a DNA Structure-specific Manner—The physical interaction detected between WRN and RAD52 suggested that the two proteins might modulate their respective catalytic activities. WRN possesses a 3'
5'-exonuclease activity that weakly hydrolyzes the 3'-recessed strand of a partial duplex DNA substrate (Table II, substrate 1) (4). Using this substrate, we observed that RAD52 inhibited not only WRN exonuclease but also the 3'5'-exonuclease activity of the E. coli DNA polymerase I Klenow fragment (data not shown). E. coli exonuclease III activity has also been reported to be inhibited by RAD52 (54), indicating that protection of DNA ends from exonuclease attack is a general property of RAD52 that is likely achieved by binding and blocking exonuclease access to the substrate rather than inhibition accomplished through specific protein-protein interactions.
WRN also possesses a 3'5' helicase activity that unwinds partial duplexes containing a 3'-overhang (5) (Table II, substrate 2) (Fig. 5A, lane 3). Under the present conditions, RAD52 alone was unable to separate the substrate into its composite strands (Fig. 5A, lane 10). We analyzed WRN helicase activity as a function of increasing RAD52 using WRN concentrations unwinding typically 15–40% of the substrate. When the substrate was added to reactions containing both proteins, unwinding was initially inhibited at sub-equimolar concentrations of RAD52, whereas, at or very close to equimolar ratios of RAD52·WRN monomers (1:1 to 1:1.6), a slight (1.5- to 2-fold) but reproducible stimulation of substrate unwinding was observed (Fig. 5, A and B). Once RAD52 concentrations exceeded that of WRN, unwinding was again inhibited as a function of RAD52 until it leveled off to 25% of the original activity observed in the absence of RAD52 (reduced to <10%). The reason for this inhibition is unclear. However, because RAD52 exhibits a higher affinity for single strand DNA compared with double strand DNA (54), the protein may bind to the 3'-single strand region, effectively blocking WRN access. Indeed, WRN unwinding of a similar partial duplex was inhibited by the presence of a streptavidin complex bound to biotin conjugated to the 3'-tail six nucleotides upstream of the single strand DNA/double strand DNA junction (55). Alternatively, at the higher concentrations, RAD52 homologous strand pairing activity may effectively re-anneal any substrate separated by WRN. The curves in Fig. 5B are sigmoidal, suggesting that stimulation is dependent on optimal cooperative interactions between the two proteins, possibly obtainable only at equimolar concentrations. As a control, helicase modulation experiments were performed using the S. cerevisiae RecQ homologue, Sgs1p. Unwinding efficiencies similar to WRN were obtained using 10-fold less Sgs1p (Fig. 5C, inset, lane 3). Co-incubation of invariant Sgs1p typically separating 50–60% of the substrate and increasing RAD52 concentrations prior to substrate addition inhibited unwinding. Unlike the situation with WRN, inhibition was observed at equimolar protein concentrations (125 pM) (Fig. 5C). Therefore, the stimulatory effect of RAD52 on partial duplex unwinding appears to be WRN-specific.
|
Because the stimulation of partial duplex unwinding by RAD52 was not dramatic, thereby calling into question the appropriateness of this substrate for measuring modulation effects, we examined possible effects of RAD52 on WRN helicase activity using structurally diverse DNA substrates. Using conditions in which stimulation had been observed with the partial duple, RAD52 did not promote unwinding on a blunt homoduplex or on otherwise homoduplex substrates containing internal (4 bp) or terminal (2 bp) mismatches (Table II; substrates 3, 4, and 5, respectively) (data not shown). However, unwinding of a blunt duplex containing a central 12-bp sequence heterology (bubble) (Table II, substrate 6) was significantly stimulated (4- to 5-fold) in the presence of increasing RAD52 at WRN concentrations that alone unwound 
10% of the substrate (Fig. 6A). As was observed with the partial duplex substrate (Fig. 5B), RAD52 enhancement of unwinding reached a maximum at nearly equimolar ratios of RAD52·WRN monomers (i.e. typically 1.3–1.7). However, unlike the situation with the partial duplex substrate, no unwinding inhibition was observed at higher RAD52 concentrations, suggesting that the dynamics between the proteins differ depending on the structure of the substrate, a phenomenon that has been described previously for WRN (3). Furthermore, unwinding inhibition due to substrate re-annealing may be abolished with this substrate due to the inability of RAD52 to efficiency pair oligonucleotides containing heterologous sequences (45). Neither RAD52 alone (Fig. 6A, inset, lane 12) nor substituting the non-hydrolysable analogue ATPS for ATP in reaction mixtures containing both WRN and RAD52 (data not shown) resulted in detectable strand separation. These results indicate that stimulation by RAD52 was dependent on the ATP-hydrolyzing activity of WRN. Typical curves of substrate unwinding as a function of RAD52 were hyperbolic (Fig. 6A), suggesting that enhancement resulted from a specific and biologically relevant interaction (5). Importantly, Sgs1p-dependent unwinding of the bubble substrate was inhibited at all quantities of RAD52 tested (Fig. 6B) demonstrating RAD52-dependent enhancement of bubble substrate unwinding reaction to be WRN-specific.
|
Four-way DNA structures, such as Holliday junctions and chicken feet, can arise as intermediates during RAD51/RAD52-dependent HR repair of double strand breaks introduced by DNA-damaging agents (i.e. gamma rays) (11), or as a consequence of replication fork reversal upon the attenuation of fork advancement (56, 57). WRN unwinds synthetic four-way junctions into single strands in vitro (42, 58), and the inability to resolve such structures in vivo has been suggested to contribute to the poor survival exhibited by WS cells following exposure to DNA-damaging agents (8). In addition, reversed replication forks, or the products of their processing by Holliday junction resolvases, may themselves be substrates for HR proteins (9). We, therefore, investigated the potential effects of RAD52 on the ability of WRN to unwind such structures. Fig. 6C shows that, under standard assay conditions, WRN alone was able to unwind, albeit weakly, about 10% of the four-way junction substrate (Table II, substrate 7) into single-stranded DNA (lane 3). However, the presence of RAD52 completely abolished WRN unwinding activity (Fig. 6C; lanes 4–6). Thus, RAD52 effectively inhibits WRN-mediated unwinding of a substrate mimicking a Holliday junction or a chicken foot.
WRN Stimulates RAD52 D-loop-forming Activity—RAD52 catalyzes annealing between single-stranded oligonucleotide and homologous DNA sequences in double-stranded plasmid present in equimolar amounts (45) (Fig. 7A, lane 3). The resulting structure is thought to resemble a displacement loop (D-loop), which is a potential substrate for WRN activity (59, 60), and a possible intermediate during recombinational repair processes. WRN alone exhibited no homologous pairing activity (Fig. 7A, lane 2) and annealing catalyzed by RAD52 was unaffected by the inclusion or exclusion of ATP in the reaction mixtures (data not shown). The addition of WRN to reactions containing equimolar concentrations of pUC19 and oligonucleotide 4, which shares complete homology with pUC19 sequences (Tables II (substrate 8) and III), preincubated with RAD52 enhanced annealing at molar ratios of RAD52·WRN monomers of typically 1.3 by 
2- to 3-fold (Fig. 7, A and B). In the presence of excess WRN, strand pairing decreased back to levels observed with RAD52 alone, possibly due to helicase unwinding and/or exonuclease degradation of the annealed structure (60). In addition, neither stimulation nor inhibition of RAD52-dependent strand annealing was observed when ATP was omitted from the reactions containing WRN (data not shown), indicating that strand annealing modulation was dependent on WRN catalytic activity. Importantly, WRN had no effect on the identical strand annealing reaction carried out by E. coli RecA recombinase (Refs. 14 and 45 and data not shown). These results indicate that modulation of strand annealing activity by WRN was specific for RAD52.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
RAD52 inhibition of WRN unwinding of the four-way junction substrate is similar to inhibition caused by p53 (61), which also interacts physically with WRN (62). Yet the observation that expression of a Holliday junction resolvase rescues both the recombination defect and cell survival following DNA damage in WS cells strongly points to four-way junctions being substrates for WRN in vivo (8). Taken together, these results suggest that WRN activity on such structures could be tightly regulated in the presence of specific proteins. Indeed, the paradigm member of the RecQ helicase family, E. coli RecQ, functions both to initiate joint molecule formation and to disrupt joint molecules formed during illegitimate recombination (63). Similar to WRN, these activities may be differentially modulated in the presence of specific proteins.
Results from the yeast two-hybrid experiments indicate that RAD52 associates with the C terminus of WRN. Furthermore, far Western analysis showed that WRN binds to both RAD52 monomers and higher molecular weight (i.e. heptamers) complexes. WRN, in turn, is thought to self-associate into trimers or hexamers, with the latter being the DNA-bound form (64). Although far Western experiments only show that RAD52 binds to WRN monomers, it does not exclude that the two proteins interact with each other in their respective oligomeric states. Indeed, biochemical analyses from the present study indicated that the most dramatic modulatory effects occurred when the two proteins were present at practically equimolar concentrations as determined on a monomer:monomer basis. Interestingly, re-determining molar ratios based on WRN hexamers:RAD52 heptamers yields values that are also at or very close to stoichiometry. Furthermore, in addition to work describing WRN self-association into trimers in solution (64), RAD52 oligomeric rings have been reported to self-associate into biologically relevant, higher order structures (65). It is, therefore, possible that the association of different and specific WRN·RAD52 multimeric complexes may be important during the various steps associated with repairing single strand gaps or DSBs under particular biological circumstances.
Endogenous WRN and RAD51 have been demonstrated to partially co-localize in response to DNA-damaging agents to foci identified to be stalled replication forks (48). Furthermore, DNA damage causes fluorescently tagged mammalian RAD51 and RAD52 proteins to co-localize in nuclear foci (49, 50). Our FRET experiments clearly show that RAD52 and WRN are present in a complex that appears to assemble at replication forks in response to DNA-damaging agents, thereby implicating, for the first time, human RAD52 in replication fork restoration. Moreover, these data, together with the observation that WRN co-precipitates from human cell extracts with both RAD52 (this study) and RAD51,2 strongly suggest that WRN, RAD52, and RAD51, possibly together with additional proteins such as RPA (replication protein A) (58, 48), participate in a complex during HR-mediated re-establishment of blocked replication forks.
In addition to HR, WRN has also been associated with NHEJ due to its ability to interact physically and functionally with NHEJ proteins (Ref. 3 and references therein). Furthermore, vertebrate organisms with defects in certain NHEJ components exhibit premature aging (66) as well as an increased tendency to develop soft tissue sarcomas harboring chromosomal translocations, amplifications, and deletions (67), features classic to WS (6). Traditionally, HR and NHEJ have been regarded as mutually exclusive processes involved in the repair of DNA strand breaks arising during different cell cycle stages (68, 69). However, biochemical analyses have suggested that, in some situations, the two pathways may not function as mutually exclusive processes but, rather, may compete with one another. Both the Ku heterodimer and RAD52 are end-binding proteins, and this property suggests that they may vie for the DNA substrate, thereby channeling repair into their respective pathways (54). Thus, WRN could be recruited to a particular recombination pathway depending on which protein "wins." Indeed, WRN is recruited to DNA by Ku (70, 71), and experimental evidence suggests that RAD52 may also recruit WRN to DNA.3 Should such a competition occur, the outcome could have important, if not dire, consequences for the cell, particularly in light of the observation that a defect in one pathway appears to shift repair to the other process (i.e. 72). Thus, as suggested, if the defect associated with WS reflects deficient HR rather than NHEJ (8), some of the observed chromosome instability could result from a shift of repair toward the more error-prone NHEJ pathway. On the other hand, the WS cellular phenotype exhibits aspects of both HR and NHEJ mutant cells suggesting rather that a lack of WRN gives rise to defects in both pathways. Interestingly, recent studies have indicated that HR and NHEJ mechanisms in eukaryotes in fact have overlapping roles and in some instances, may even be coupled (68, 69). Indeed, both RAD51-dependent HR and NHEJ have been demonstrated to participate in replication fork rescue, although they may perform slightly different tasks (9, 73). The dynamics of WRN association with either pathway during replication fork re-establishment remain to be determined.
Recombination processes, in addition to replication fork recovery and repair of DNA discontinuities, have also been associated with telomere metabolism (74–77) and transcription (78, 79). Indeed, RAD52 has recently been shown to associate with transcription factors and to function in transcriptional regulation (80). If recombination is an important component of these pathways, then it follows that a recombination deficit can account for the defective telomere metabolism and transcription also observed in WS cells (3, 81). The proposed function of WRN in strand annealing and joint molecule resolution associated with fork recovery could also be pertinent to recombination associated with both telomere maintenance (82) and transcription. Further work is anticipated to reveal important connections between recombination, WRN activity, and aspects of DNA metabolism important to the maintenance of genome stability.
|
FOOTNOTES |
|---|
* This work was supported by the Research Council of Norway, by the Norwegian Cancer Society, and by the European Union (Contract QLRT 1999-02002). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 
Present address: Biotechnology Research Institute, National Research Council of Canada, 6100 Royalmount Ave., Montreal, Quebec H4P 2R2, Canada.
** To whom correspondence should be addressed. Tel.: 472-307-4059; Fax: 472-307-4061; E-mail: e.c.seeberg{at}labmed.uio.no.
1 The abbreviations used are: WS, Werner syndrome; DSBs, double strand breaks; NHEJ, non-homologous end-joining; HR, homologous recombination; FRET, fluorescence resonance energy transfer; MMC, mitomycin C; AD, activation domain; BD, binding domain; BSA, bovine serum albumin; PMSF, phenylmethylsulfonyl fluoride; DTT, dithio-threitol; PCNA, proliferating cell nuclear antigen; NFRET, normalized FRET; D-loop, displacement loop; ATPS, adenosine-5'-o-(3-thiotriphosphate); IgGHC, IgG heavy chain; ECFP, enhanced cyano fluorescent protein; EYFP, enhanced yellow fluorescent protein.
2 K. Baynton, M. Otterlei, M. Bjørås, C. von Kobbe, V. A. Bohr, and E. Seeberg, unpublished data.
3 K. Baynton, unpublished data.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
