The human Rothmund-Thomson syndrome gene product, RECQL4, localizes to distinct nuclear foci that coincide with proteins involved in the maintenance of genome stability

Genome instability is thought to play a major role in the development and progression of cancer, and has also been implicated in the aging process. One important family of proteins required to maintain genome stability is the RecQ family of DNA helicases (Hickson, 2003). This highly-conserved protein family includes Escherichia coli RecQ (Nakayama et al., 1984), Saccharomyces cerevisiae Sgs1p (Gangloff et al., 1994), Schizosaccharomyces pombe Rqh1p (Enoch et al., 1992), and at least five human homologs named RECQL1 (Puranam and Blackshear, 1994; Seki et al., 1994), WRN (Yu et al., 1996), BLM (Ellis et al., 1995), RECQL4 and RECQL5 (Kitao et al., 1998). Common to all RecQ family members is a helicase domain that comprises seven highly conserved motifs of the DExH-box family that are found in many DNA and RNA helicases (Singleton and Wigley, 2002). During the past decade, there has been increasing interest in the RecQ family of DNA helicases because a group of rare, autosomal recessive human disorders has been linked to mutations in some of its members. These disorders include Werner's syndrome (WS), the premature aging disease caused by mutations in WRN gene (Yu et al., 1996); Bloom's syndrome (BS), a cancer predisposition disease caused by mutations in BLM gene (Ellis et al., 1995), and a subset of cases of RTS characterized by mutations in the RECQL4 gene (Kitao et al., 1999b).

RTS displays heterogeneous clinical profiles that include growth-deficiency, photosensitivity with poikiloderma, cataracts, early graying and loss of hair, as well as the early development of cancers of many types, in particular osteosarcoma (Lindor et al., 2000). The gene mutated in a subset of RTS patients, RECQL4, contains 21 exons spanning only 6 kb of genomic DNA sequence and is located on human chromosome 8q24.3 (Kitao et al., 1999a). In addition to RTS, mutations in the RECQL4 gene can also cause an RTS-similar disease termed RAPADILINO syndrome (Siitonen et al., 2003). The latter is an autosomal recessive disorder characterized by short stature, radial ray defects and other malformations, as well as infantile diarrhoea, but not by a significant cancer risk (Jam et al., 1999). The most common mutations of the RECQL4 gene in RAPADILINO patients represent in-frame deletions of exon 7 (Siitonen et al., 2003). On the other hand, most alternations of RECQL4 found in RTS patients represent nonsense or frameshift mutations, resulting in a truncated polypeptide and map to the helicase domain (Kitao et al., 1999b; Wang et al., 2003). Somatic cells isolated from RTS individuals are sensitive to ionizing radiation (IR) (Vennos and James, 1995) and show genomic instability, including trisomy, aneuploidy and chromosomal rearrangements (Der Kaloustian et al., 1990). In agreement with this notion, targeted disruption of RECQL4 in mice results in abnormalities similar to those in RTS patients (Hoki et al., 2003). During G1 arrest, the promoter of the RECQL4 gene is repressed by p53 that forms a complex with the transcription factor SP1 in concert with histone deacetlyase 1 (HDAC1) (Sengupta et al., 2005). In human cells, RECQL4 is associated with ubiquitin ligases UBR1 and UBR2, thus indicating a potential role of RECQL4 in the N-end rule pathway (Yin et al., 2004).

Numerous studies on Sgs1p, Rqh1, BLM and WRN have revealed that RecQ helicases are involved in the maintenance of genome stability by functioning at the interface between DNA replication, recombination and repair (Hickson, 2003). However, the biochemical properties, interaction partners and subcellular localization of RECQL4 have to date been largely unexplored. In this study, we have carried out a cellular localization analysis of the human RECQL4 protein. To achieve this goal, we have raised two different anti-RECQL4 antibodies that recognize a RECQL4-specific band from human cell extracts. We show that RECQL4 forms nuclear foci in several human cell lines, the number of which does not significantly change after induction of different types of DNA damages. Importantly, downregulation of RECQL4 by siRNA causes reduction in RECQL4 nuclear foci formation. The RECQL4 foci coincide with those formed by PML nuclear bodies and the number of colocalizing foci slightly decreases after induction of DNA double-strand breaks (DSBs). In addition, RECQL4 forms a complex with the human Rad51 protein and colocalizes with Rad51 and regions of single-stranded DNA (ssDNA) upon induction of DNA DSBs. These findings suggest a role for RECQL4 in the repair of DSBs by homologous recombination and indicate a novel function for RECQL4 in human cells.

The HeLa, 293T and WI38/VA13 cells were from the American Type Culture Collection (ATCC), AG18371 cells from Coriell and the primary skin fibroblasts were a kind gift from Silvio Hemmi, University of Zurich, Switzerland. Cells were cultured in Dulbecco's modified Eagle's medium (Life Technologies) and 10% FBS. For some experiments cells were exposed to 10 μM etoposide for 1 hour at 37°C, washed, given drug-free medium, and collected 10-12 hours later. For detection of ssDNA after damage, cells were grown on slides in medium containing 10 μg/ml BrdU, exposed to 10 μM etoposide for 1 hour in BrdU-free medium and fixed after 10 hours. For HU treatment cells were incubated with 10 mM HU for 24 hours, UV treatment was performed with 50 J/m² at 254 nm, cells were fixed 16 hours after the treatment; for ionizing-radiation treatment cells were exposed to 5 Gy and fixed 4 hours later. For the indirect immunofluorescence assays, cells growing on glass slides were incubated with 0.5% Triton X-100 in PBS for 5 minutes, fixed with 2% paraformaldehyde in PBS (20 minutes at RT), and permeabilized with 0.5% Triton X-100 in PBS (20 minutes at RT). After the blocking step (0.1% Tween-20, 5% BSA in PBS, 20 minutes at RT), slides were incubated with a mixture of different rabbit and mouse antibodies. Rabbit antibodies were detected with Cy3-conjugated goat anti rabbit IgG (Jackson Laboratories; 1:200 in blocking buffer), and mouse antibodies were detected with FITC-conjugated goat anti mouse IgG (Jackson Laboratories; 1:100 in blocking buffer). To visualize nuclear DNA, slides were incubated with DAPI (0.4 μg/ml). After washing, slides were mounted in Vectashield (Vector Laboratories) and viewed under confocal microscope Leica TCS 4D. Images were processed by Imaris software. For statistical analysis a minimum of 200 cells was counted in two independent experiments.

The rabbit polyclonal RECQL4 N-t and RECQL4 C-t antisera were raised against recombinant human RECQL4 fragments comprising amino acid residues 359 to 478 and 911 to 1208, respectively. For that, the corresponding cDNA fragments were cloned in pET-30 (RecQL4 N-t) or pET-28 (RECQL4 C-t) expression vectors (Novagen) and recombinant proteins fused to a histidine tag were purified using Ni-NTA resin (Qiagen) following the manufacturer's instructions.

RECQL4 N-t and C-t antibodies were affinity-purified using hexahistidine-tagged RECQL4 N-t and C-t recombinant proteins as an affinity matrix, respectively. The purified proteins were fractionated by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were stained with Ponceau S, and the corresponding 6xHis-RECQL4 band was excised. The strips were blocked for 1 hour with 2% bovine serum albumin in TBS (10 mM Tris, pH 7.4, 150 mM NaCl) containing 0.1% Tween 20. The strips were then incubated overnight with rabbit antiserum and washed with TBS containing 0.1% Tween 20 and 1 M NaCl. Anti-RECQL4 antibodies were eluted by incubation in 100 mM glycine pH 2.3, for 5 minutes. The eluate was then immediately neutralized with 1 M Tris-HCl pH 8.8.

In immunofluorescence experiments, anti-PML antibody (Santa Cruz, PG-M3) was used in a 1:50 dilution, anti-Rad51 antibody (abcam, ab1837) in a 1:50 dilution, anti-BrdU antibody (Pharmingen, 3D4) in a 1:50 dilution, anti-RECQL4 N-t antibody in a 1:200 dilution. For western blots, anti-RECQL4 N-t and anti RECQL4 C-t antibodies were used in a 1:2000 dilution, anti-tubulin antibody (Santa Cruz, TU-02) in a 1:1000 dilution, anti-PARP1 antibody (Petrilli et al., 2004) in a 1:2000 dilution, anti-BLM antibody (Wu et al., 2000) in a 1:1500 dilution, anti-GRB2 antibody (BD Transduction, G10111) in a 1:2000 dilution, anti-FLAG antibody (Sigma, M2) in a 1:2000 dilution and anti-Rad51 antibody (Tan et al., 1999) in a dilution 1:20,000.

Total cell extracts were prepared from exponentially growing HeLa cells by lysis in lysis buffer (50 mM Hepes, pH 7.6; 500 mM NaCl; 1% Triton X-100, protease inhibitors) for 15 minutes with rolling at 4°C and centrifugation at 14,000 rpm for 15 minutes at 4°C. Nuclear extracts were prepared from HeLa cells, which were washed in PBS and re-suspended in buffer A (10 mM Hepes, pH 7.9; 1.5 mM MgCl₂; 10 mM KCl; protease inhibitors) and then centrifuged for 1 minute at 2000 rpm. The cell pellet was re-suspended in buffer A+ (buffer A; 0.1% Nonidet P 40) for 5 minutes on ice and centrifuged at 10,000 rpm, for 5 minutes. The supernatant contained cytosolic extract. The pellet was washed in buffer A and then re-suspended in buffer C (20 mM Hepes, pH 7.9; 420 mM NaCl; 1.5 mM MgCl₂; 25% glycerol, protease inhibitors) for 15 minutes, with rolling at 4°C. The samples were then centrifuged at 14,000 rpm, for 5 minutes. The supernatant contained nuclear extract and the NaCl concentration was adjusted to 120 mM with buffer D (20 mM Hepes, pH 7.9, protease inhibitors).

293T cells were transiently transfected with p3xFLAG-myc-CMV-23 (Sigma) expressing RECQL4 or vector alone. Preparation of nuclear extracts was described in previous section. Aliquots (120 μg) of nuclear extracts were incubated with 3 μg of anti-FLAG antibody coupled to magnetic tosylactivated Dynabeads (Dynal Biotech, M-280) according to manufacturer's instructions in immunoprecipitation buffer (20 mM Hepes, pH 7.5; 120 mM NaCl; 5 mM MgCl₂; 0.1% Nonidet P 40; protease inhibitors) at 4°C overnight. As a control, nuclear extracts from untransfected cells were used. The beads were washed three times in immunoprecipitation buffer before the protein complexes bound to the beads were eluted and analysed by sodium dodecyl sulfate-polyacrylamide (SDS) gel electrophoresis. A 12 μg aliquot of nuclear extract was used as input control. For the reverse immunoprecipitation assay, an aliquot of nuclear extracts (120 μg) was incubated with 3 μl (approximately 2 μg) of polyclonal anti-Rad51 antibody (kind gift of Roland Kanaar, Erasmas University, Rotterdam, The Netherlands) in immunoprecipitation buffer for 3 hours, to which 20 μl of protein G sepharose 4 fast flow (Amersham Biosciences) beads were added and incubated for an additional for 2 hours. As a control, nuclear extracts were incubated with 1.2 μg of rabbit IgG (Santa Cruz). The beads were washed three times in immunoprecipitation buffer before the protein complexes bound to the beads were eluted and analysed by SDS gel electrophoresis.

For short interfering RNA (siRNA) experiments, RECQL4-1 5′-aacuucugaucccuggugagu-3′ and control-1 5′-aaucuagcucuccuguagggu-3′ with dTdT overhang (Qiagen, Operon) were used. The siRNAs were transfected in HeLa cells using a 200 nM siRNA concentration with Oligofectamine (Invitrogen) according to manufacturer's instruction and harvested after 24 hours.

Antibodies against N- and C-termini of RECQL4 were raised in rabbits using chimeric proteins consisting of 6xHis-tag fused to residues 359-478 and 911-1208 of RecQL4. These regions of RECQL4 were chosen because they show no sequence homology to other human RecQ helicases. These sera, designated anti-RECQL4 N-t and anti-RECQL4 C-t, respectively, were affinity-purified as described in Materials and Methods. In western blots, both sera recognized their protein fragments used for immunization (data not shown), and a major protein band in HeLa nuclear, cytosolic and total cell extracts with a molecular mass of approximately 150 kDa (Fig. 1A). To control our cell fractionation experiments, we examined the distribution of the endogenous human nuclear proteins PARP1 and BLM as well as a cytosolic protein GRB2. As expected, a band corresponding to PARP1 and BLM was detected in the nuclear extract only, whereas the band corresponding to GRB2 was detected only in the cytosolic extract (Fig. 1A, three lower panels).

In addition, both RECQL4 antisera also recognized FLAG-epitope-tagged RECQL4, expressed in 293T cells, with the expected molecular mass (Fig. 1B). No such protein was detected in HeLa nuclear and total cell extracts with preimmune serum (data not shown) or cell in extracts derived from the RTS-patient cell line AG18371, using both of the anti-RECQL4 sera (Fig. 1A). Likewise, both RECQL4 antibodies did not cross-react with purified recombinant BLM and WRN helicases (data not shown). These results, and the results from the siRNA experiments (discussed below), show that both antisera are specific for the RECQL4 protein.

Next, we used the affinity purified anti-RECQL4 N-t antiserum to visualize the in situ distribution of RECQL4 in human cells. Indirect immunofluorescence staining analysis by confocal microscopy of exponentially growing HeLa, WI38/VA13 and primary skin fibroblast cell lines revealed a punctate nuclear staining in all three cell lines, which was observed in the majority of the cells (Fig. 2A-C). No cytoplasmic staining was evident in any of the examined cells, which is contradictory to our western blot results. This can be explained by the difference in sample preparation and detection methods used in these two assays, and that the protocol used for our immunofluorescence assays presumably removes the RECQL4 protein - which is not bound to the nuclear matrix. The calculated average number of RECQL4 foci formed per cell is 25. Furthermore, no specific fluorescence was seen in the cells with pre-immune serum (Fig. 2D).

To further confirm the specificity of RECQL4 nuclear foci staining, we silenced RECQL4 using RNA interference. We transiently transfected HeLa cells with siRNA RECQL4-1 directed against the RECQL4 mRNA, and control-1 siRNA containing a scrambled sequence of RECQL4-1. We first examined the efficiency of silencing of RECQL4 by immunoblotting. As shown in Fig. 3A, silencing of RECQL4 occurred only in cells transfected with RECQL4-1 siRNA, in which approximately a 70% decrease in the level of endogenous RECQL4 was detected compared with the cells transfected with a control-1 siRNA. An even more pronounced effect was observed after immunofluorescence staining analysis (Fig. 3B). We detected a significant reduction of RECQL4 nuclear foci in the cells transfected with siRNA RECQL4-1, which was not the case for the cells treated with the control-1 siRNA (Fig. 3B,C).

We also examined whether the RECQL4 nuclear foci staining changes after induction of DNA damage. For that purpose, HeLa and WI38/VA13 human fibroblasts were treated with different DNA damaging agents including ionizing radiation (IR), which produces DSBs; ultraviolet light (UV), which generates predominantly pyrimidine dimers and 6-4 photoproducts; and hydroxyurea (HU), an inhibitor of DNA replication. As shown in supplementary material Fig. S1, indirect immunofluorescence analysis showed that there was no significant change in the number of RECQL4 foci after these treatments.

Previous reports have identified a dynamic association of BLM with promyelotic leukaemia protein (PML) (Bischof et al., 2001; Zhong et al., 1999). PML functions as a tumor suppressor, and localizes to nuclear foci termed PML nuclear bodies (PML NBs) (Hodges et al., 1998). It has been suggested that these localization sites function as protein-storage sites, sites where multi-subunit complexes form and post-translational modification of regulatory factors occur, and that are modified in response to a variety of stresses including oncogenes (Dellaire and Bazett-Jones, 2004), and DNA breaks (Carbone et al., 2002). We have therefore analyzed whether the degree to which RECQL4 foci colocalized with PML in exponentially growing, untreated HeLa, WI38/VA13 and primary skin fibroblast cells. In these cells, we observed a partial colocalization between RECQL4 and PML (Fig. 4, upper panel). On average, 14% of RECQL4 foci colocalized with PML foci and 27% of PML foci colocalized with RECQL4 foci, an observation seen in 90% of cells. We then wanted to test whether the number of colocalizing RECQL4 and PML foci changes after treatment of cells with DNA-damaging agents. Upon treatment with a DNA-DSB-inducing agent, etoposide, the number of colocalizing foci was slightly reduced (Fig. 4, lower panel). In 90% of treated cells, 10% of RECQL4 foci colocalized with PML foci and 18% of PML foci colocalized with RECQL4 foci. Collectivelly, these data indicate that at least a portion of RECQL4 foci is associated with PML NBs and that, after induction of DNA DSBs, a minor portion of RECQL4 foci relocalizes from PML NBs.

We next wanted to determine whether RECQL4 is present at regions of ssDNA after the induction of DSBs. For that purpose, 5-bromo-2′-deoxyuridine (BrdU), was used to visualize sites of ssDNA that are often associated with DSB repair. BrdU is inaccessible to an anti-BrdU antibody when present in duplex DNA, but readily accessible to the antibody when the DNA is denatured or single-stranded (Raderschall et al., 1999). HeLa cells that were grown in medium containing BrdU, then fixed with paraformaldehyde and stained under non-denaturing conditions, showed no significant staining, confirming that the antibody does not detect BrdU in duplex DNA (Fig. 5A, upper panel). The same cells, however, showed the expected pattern of RECQL4 foci, indicating that BrdU does not perturb RECQL4 localization. After exposing BrdU-labelled cells to etoposide, BrdU foci appeared in the majority of the cells (Fig. 5A, lower panel). These foci formed because of DNA DSBs, where they presumably identify sites of repair. To determine whether RECQL4 localizes with BrdU, we co-stained for RECQL4 and BrdU 10 to 12 hours after etoposide treatment. We could detect partial colocalization between RECQL4 and regions of ssDNA as determined by anti-BrdU staining (Fig. 5A, lower panel). There were 9% of RECQL4 foci colocalizing with BrdU foci and 18% of BrdU foci colocalizing with RECQL4 foci. The RECQL4/BrdU colocalization was evident in 46% of cells. Similar results were obtained after cell treatment with IR (data not shown). These results suggest that a portion of RECQL4 foci can associate with regions of ssDNA after exposure to DSB-inducing agent, and might represent sites of ongoing DNA repair.

It has been previously reported that Rad51 foci, which are important for DNA repair by homologous recombination, accumulate at sites of ssDNA after induction of DSBs (Raderschall et al., 1999). The colocalization of RECQL4 and regions of ssDNA after induction of DSBs prompted us to test whether RECQL4 is a component of Rad51 foci after the same treatment. In exponentially growing, untreated asynchronous WI38/VA13 cells, no Rad51 foci were observed (Fig. 5B, upper panel). This is in agreement with previous studies, which showed that Rad51 foci are detectable only in S phase and/or after induction of DSBs (Tashiro et al., 1996; Thompson and Schild, 2001). However, after we treated the cells with etoposide, the percentage of cells with focally concentrated Rad51 increased. We found that in 43% of these cells, 29% of Rad51 foci colocalized with RECQL4 foci and 10% of RECQL4 foci colocalized with Rad51 foci, indicating that at least a portion of RECQL4 nuclear foci is recruited to sites of DNA-damage-induced ssDNA (Fig. 5B, lower panel). Similar results were obtained with HeLa cells and primary skin fibroblast cells after exposure to etoposide (data not shown). Additionally, in etoposide-treated HeLa cells transfected with RECQL4-1 siRNA, but not in cells transfected with the control siRNA, there was a significant reduction in the formation of RECQL4 foci and thus no colocalization of RECQL4 with Rad51 (supplementary material Fig. S2).

We next analysed whether RECQL4 and Rad51 interact in human cells using a co-immunoprecipitation assay. To that end, the FLAG-RECQL4 was transiently transfected into 293T cells. Nuclear extracts were then immunoprecipitated with anti-FLAG resin and analysed by immunoblotting using the anti-Rad51 antibody. As can be seen in the left panel of Fig. 5C, endogenous Rad51 was efficiently co-precipitated by the anti-FLAG antibody (lane 3), but was not detected when the control IgG antibody was used (lane 2). The reciprocal co-immunoprecipitation experiment was also carried out with an anti-Rad51 polyclonal antibody, to immunoprecipitate FLAG-RECQL4 from 293T nuclear extracts. As shown in the right panel of Fig. 5C, RECQL4 could be specifically co-immunoprecipitated with endogenous anti-Rad51 (lane 6), but not with the control IgG antibody (lane 5). Taken together, these data indicate that RECQL4 and Rad51 form a complex in human cells and that, in response to the induction of DSBs, a portion of RECQL4 and Rad51 nuclear foci colocalize to sites that represent sites of undergoing repair by homologous recombination.

BRCA1, a multifunctional protein with proposed roles in the repair of DSBs by homologous recombination, transcriptional activation, cell-cycle regulation and apoptosis (Narod and Foulkes, 2004), has been shown to colocalize with Rad51 in S-phase foci (Scully et al., 1997) and IR-inducible foci (Bhattacharyya et al., 2000). We therefore wanted to determine whether RECQL4 protein colocalizes with BRCA1, by performing double immunolabeling in untreated HeLa cells and cells treated with 10 μM etoposide. In contrast to the partial RECQL4/Rad51 colocalization, a very weak colocalization of RECQL4 with BRCA1 was observed (Fig. 5D). Namely, 2% of RECQL4 foci colocalized with BRCA1 foci and 5% of BRCA1 foci colocalized with RECQL4 foci, an observation made in only 11% of cells. We conclude from this result that RECQL4 foci differ from BLM foci, which colocalize with the BRCA1 protein (Jiao et al., 2004; Wang et al., 2000).

Numerous genetic, biochemical and cell biological studies in yeast and human cells have suggested that the members of the RecQ helicase family play a pivotal role in maintaining genome stability (Hickson, 2003). Along with this notion, mutations in the RECQL4 gene have been shown to cause some cases of the Rothmund-Thomson and RAPADILINO syndromes, which have severe physiological consequences in humans, the most prominent of which are cancer susceptibility and premature aging. At the cellular level, defects in RECQL4 cause genome instability particularly trisomy, aneuploidy, deletions, translocations and high frequencies of chromosomal rearrangements. Despite this knowledge, the molecular function of RECQL4, as well as the cellular pathways in which it is involved, remain poorly understood. In the current study, we have shown for the first time that endogenous RECQL4 localizes to distinct nuclear foci in various human cell lines. This nuclear foci staining pattern is specific for RECQL4 because silencing of RECQL4 through RNA interference markedly decreases the RECQL4 nuclear foci formation. In contrast to previous studies involving BLM and WRN helicases, which have shown their presence in the nucleus (Bischof et al., 2001; Jiao et al., 2004; Pedrazzi et al., 2001; Sakamoto et al., 2001; Yankiwski et al., 2000), our cellular fractionation experiments showed that, although a majority of RECQL4 is present in the nucleus, a small amount of it can be found in the cytosolic fraction of HeLa cell extracts. A similar result has also been recently reported by Yin and colleagues, who showed that the greater portion of RECQL4 protein is present in the cytosol compared to the nucleus of transformed human cells (Yin et al., 2004). At the present time, the precise role of such `cytosolic RECQL4' is not known and remains to be determined. Our data that RECQL4 forms nuclear foci seemingly contrasts with the data of Yin and colleagues, who reported that RECQL4 is present in both the cytoplasm and the nucleus of human cells, with a punctate and uniform distribution of RECQL4 in both compartments (Yin et al., 2004). These differences are probably the result of different cell fixation approaches and antibodies that have been employed in these two studies.

Furthermore, we have provided evidence that the number of RECQL4 foci does not significantly change upon cellular treatment with different DNA damaging agents including IR, UV light and HU. Interestingly, these data differ from those obtained for BLM and WRN helicases, and may thus suggest that RECQL4 has a different function in human cells. Namely, previous studies have shown that BLM assembles into nuclear foci specifically in response to agents that cause DNA DSBs (Bischof et al., 2001; Davalos and Campisi, 2003) and block replication (Jiao et al., 2004; Sengupta et al., 2004). However, WRN was shown to traffic between the nucleolus and nucleoplasm, and to form nuclear foci in response to treating cells with various DNA-damaging agents (Gray et al., 1998; Sakamoto et al., 2001).

We have also found that a portion of RECQL4 foci colocalizes with PML NBs in untreated human cells. PML NBs have been shown to be the sites of BLM, RAD51, MRE11 and p53 assembly during late S-G2 phase and following the induction of DSBs (Bischof et al., 2001; Carbone et al., 2002). It has therefore been suggested that these foci represent sites of presumptive homologous recombinational repair after induction of DSBs (Bischof et al., 2001). Given the fact that a portion of RECQL4 foci colocalizes with PML NBs, our findings suggest that RECQL4 plays an active role in DSB repair. Furthermore, we have observed a slight decrease in the number of RECQL4- and PML-colocalizing foci after treatment of cells with the DSB-inducing agent etoposide. Similarly to this observation, it has recently been reported that, BLM helicase relocalizes from PML NBs after replication stall due to HU treatment (Davalos et al., 2004). Thus, our results also suggest that RECQL4 leaves PML foci and, after induction of DNA damage, relocalizes in a similar manner as BLM. In other words, there are different pools of RECQL4 foci in the cell that have a different function after induction of DNA damage.

We have also demonstrated that RECQL4 is a component of a portion of Rad51 foci that is thought to be crucial for the repair of DSBs by homologous recombination. We found that RECQL4 forms a complex with Rad51 in human cells and partially colocalizes with Rad51 and regions of ssDNA after induction of DSBs. Nuclear foci containing the Rad51 protein were shown to include other recombination proteins, such as Rad52, Rad54, the single-strand-binding-protein RPA, and the tumor suppressors BRCA1 and BRCA2 (Tarsounas et al., 2004). Given the fact that the processing of DSBs frequently involves the formation of long 3′ single-stranded regions and that Rad51 is tightly associated with these structures in vivo, which promote pairing and strand-exchange between ssDNA and homologous dsDNA (Raderschall et al., 2002; Raderschall et al., 1999), one possible interpretation of the RECQL4/Rad51 complex formation, and RECQL4/PML as well as RECQL4/Rad51 colocalization is that, these structures represent the physical sites where the repair of DSBs by homologous recombination is coordinated and monitored. In agreement with this observation it has previously been shown that cultured cells from RTS patients are sensitive to IR, which further suggests that RECQL4 might participate in DSB repair (Vennos and James, 1995). However, it should be noticed that, it remains to be determined whether RECQL4 has a direct or indirect role in DSB repair by homologous recombination. Similarly to BLM and WRN helicases, RECQL4's function could be to resolve aberant DNA structures that are formed during DNA replication and recombinational repair processes. Alternatively, RECQL4 could have a signalling function during the same processes through interactions with other proteins. Either hypothesis or a combination of the two would explain the genome instability displayed by RTS cells and the high incidence of cancer seen in RTS patients.

Our results also show that RECQL4 foci, unlike BLM foci, do not coincide with BRCA1 (Jiao et al., 2004). Previous work has revealed that BRCA1 colocalizes and resides in a so-called `BASC' (BRCA1-associated genome surveillance complex) that consists of 40 different proteins, including BLM, RAD50, ATM, RAD50/MRE11/NBS1 complex, MSH6, MSH2, MLH1 and three subunits of the replication factor C (RFC) complex (namely p140, p37 and p34) (Wang et al., 2000). BLM and BRCA1 foci do not coincide in untreated human cells; however, their colocalization is greatly enhanced in those cells that are in mid-to-late S phase or in G2 phase after treatment with HU or IR, which suggests a specific requirement for BLM/BRCA1 in replication and/or repair of late-replicating DNA (Wang et al., 2000). In this way, our results imply that RECQL4 foci are distinct from the BLM foci, suggesting that RECQL4 does not participate in the same damage response pathway as BLM.

In conclusion, we have shown that endogenous human RECQL4 localizes to distinct nuclear foci that coincide with PML NBs, Rad51 and regions of ssDNA, suggesting that RECQL4 plays a role in the repair of DSBs by homologous recombination. Further confirmation of the above mentioned results will come from genetic, biochemical and protein-protein-interaction assays performed on RECQL4. These approaches, along with the cell biological approach described in this study, will ultimately help to understand the function of RECQL4 protein in human cells and to understand how genome instability influences cancer predisposition and aging.

We thank Silvio Hemmi for the gift of primary skin fibroblast cell line, Roland Kanaar for the gift of the anti-Rad51 antibody, Paul Hassa and Michael Hottiger for the gift of the anti-PARP1 antibody, Veronique Smits for the gift of the anti-GRB2 antibody, Ian Hickson for the gift of the anti-BLM antibody and Vilhelm Bohr, Michael Fetchko, Ian Hickson, Orlando Schärer, Caroline Zbinden and other members of the Stagljar group for helpful discussions. We are grateful to the Elektronenmikroskopisches Zentrallaboratorium der Universität Zürich for providing confocal facilities for this study. I.S.'s group is financed by grants from the Union Bank of Switzerland (im Auftrag eines Kunden), Gebert Rüf Foundation, EU Grant HPRN-CT-2002-00240, Schweizer Bundesamt für Bildung und Wissenschaft BBW Nr. 01-0528, the Swiss Cancer League (OCS-01310-02-2003) and the Swiss National Science Foundation (Nr. 3100A0-100256/1). R.F. is supported by the EU Grant HPRN-CT-2002-00240 and the GEN2003-20243-C08-07 from the Spanish Ministry of Education and Science.