Little is known about the functional interaction between the Bloom's syndrome protein (BLM) and the recombinase RAD51 within cells. Using RNA interference technology, we provide the first demonstration that RAD51 acts upstream from BLM to prevent anaphase bridge formation. RAD51 downregulation was associated with an increase in the frequency of BLM-positive anaphase bridges, but not of BLM-associated ultrafine bridges. Time-lapse live microscopy analysis of anaphase bridge cells revealed that BLM promoted cell survival in the absence of Rad51. Our results directly implicate BLM in limiting the lethality associated with RAD51 deficiency through the processing of anaphase bridges resulting from the RAD51 defect. These findings provide insight into the molecular basis of some cancers possibly associated with variants of the RAD51 gene family. Mol Cancer Res; 8(3); 385–94

Bloom's syndrome displays one of the strongest known correlations between chromosomal instability and a high risk of cancer at an early age. Bloom's syndrome is caused by mutations in the BLM gene, which encodes BLM, a RecQ 3′-5′ DNA helicase (1). The hallmark of Bloom's syndrome cells is a high rate of sister chromatid exchange (SCE), generally thought to be the consequence of replication-dependent double-strand breaks. SCEs are mediated by homologous recombination (HR; refs. 2-4). BLM deficiency is associated with replication abnormalities and an increase in HR (1). In vitro, BLM unwinds DNA structures mimicking replication forks and HR intermediates. It catalyzes the regression of replication forks and works with topoisomerase IIIα, RMI1, and RMI2 to dissolve double Holliday junctions (5-10). BLM inhibits the D-loop formation catalyzed by RAD51 by displacing RAD51 from ssDNA, thereby disrupting nucleoprotein filaments (11). In vivo evidence that BLM can disrupt RAD51 polymerization has also been reported (12). Recently, SUMOylation of BLM has been shown to regulate its association with RAD51 and its function in HR-mediated repair of damaged replication forks (13). In several models, it has been proposed that BLM restarts replication after the stalling of the fork, thus preventing HR-dependent fork restart (14-16). The Bloom's syndrome phenotype also includes a significant increase in the frequency of anaphase bridges and lagging chromosomes during cell division, indicating a defect in sister chromatid separation during mitosis (17). BLM localizes to anaphase bridges and is required for their elimination (17), suggesting that BLM helps to resolve aberrant chromosome structures generated during DNA replication (17, 18). Moreover, BLM is associated with 4′,6-diamidino-2-phenylindole (DAPI)–negative ultrafine DNA bridges (UFB), which are thought to be derived from replication or recombination intermediates (17). In particular, BLM localizes to UFBs and to conventional DAPI-positive anaphase bridges associated with fragile sites and is required for their resolution (19). RAD51 downregulation significantly increases fragile-site expression under both normal conditions and replication stress (20). These data suggest that a functional relationship between BLM and RAD51 may also exist during mitosis. Using RNA interference to downregulate BLM and/or RAD51 in HeLa cells, we show here that BLM and RAD51 interact functionally in both interphase and mitotic cells. Indeed, we report an epistatic interaction between BLM and RAD51 in SCE formation. We also show that (a) RAD51 acts upstream from BLM to prevent anaphase bridge formation; (b) RAD51 downregulation is associated with an increase in the frequency of BLM-positive anaphase bridges, but not of BLM-associated UFBs; and (c) BLM downregulation increases the levels of cell death associated with RAD51 downregulation. Our results indicate that BLM acts downstream from RAD51 to resolve RAD51-mediated Holliday junctions or to rescue anaphase bridges resulting from RAD51 deficiency, probably at difficult-to-replicate DNA sequences such as fragile sites.

Cell Cultures and Transfections

HeLa cells were used, as previously described (21). HeLaV cells and HeLashBLM cells were obtained by transfecting HeLa cells with an empty pSM2 vector or a pSM2 vector encoding a short hairpin RNA (shRNA) sequence directed against BLM (Open Biosystems, clone V2HS-89234), respectively, using JetPEI reagent (Ozyme) according to the manufacturer's instructions. After 48 h, selection with 1 to 5 μg/mL puromycin (Invivogen) was applied. Individual colonies were isolated and maintained in growth medium containing 0.5 μg/mL puromycin.

HeLa cells producing H2B fused to green fluorescent protein (GFP) were kindly provided by Dr. Kevin F. Sullivan (NUI Galway, Galway, Ireland; ref. 22). For siRNA transient transfection assays, 3 × 105 to 4 × 105 cells were added to 3 mL of medium in each well of a six-well plate. Cells were transfected by incubation with siRNAs specific for BLM or RAD51 (ON-TARGETplus, SMARTpool, Dharmacon) or with negative control siRNAs (ON-TARGETplus siCONTROL Non Targeting Pool, Dharmacon) at a final concentration of 100 nmol/L (single transfections) or 200 nmol/L (cotransfections) for 24 to 72 h, in the presence of DharmaFECT I (Dharmacon), used according to the manufacturer's instructions.

Chemicals

Hydroxyurea (Sigma) was used at a final concentration of 2 mmol/L.

Assays to Assess the Efficiency of Colony Formation

Untreated cells were plated in a drug-free medium at three different densities, in triplicate, for the counting of 30 to 300 clones depending on expected survival. After 14 to 21 d of incubation, colonies were fixed and stained with methylene blue (5 g/L in 50% water and 50% methanol) and scored. Only experiments giving a linear correlation between the different dilutions were considered. Colony-forming efficiency was estimated by dividing the number of colony-forming units by the number of cells plated.

Western Blot Analysis

Cells were lysed by incubation in 1% SDS in water or in 350 mmol/L NaCl, 50 mmol/L Tris-HCl (pH 7.5), 1% NP40, 1 mmol/L NaF, and protease inhibitors (23) for 30 min on ice, sonicated, and heated. Samples equivalent to 1.25 × 105 or 2.5 × 105 cells or to 45 μg of protein were subjected to SDS-PAGE in a 5.5% or 12% SDS polyacrylamide gel or in NuPAGE Novex 4% to 12% Bis-Tris pre-cast gels (Invitrogen), respectively. The procedures used for gel electrophoresis and immunoblotting were as previously described (24).

Antibodies

All the commercial antibodies were used according to the manufacturers' specifications. The primary antibody used against BLM was ab476 (1:1,000; rabbit, Abcam). We used rabbit polyclonal antibodies against RAD51 (1:1,000; AB-1, Calbiochem) and β-actin (1:10,000; Sigma).

Horseradish peroxidase–conjugated goat anti-mouse IgG and goat anti-rabbit IgG (Santa Cruz Biotechnology) were used at a dilution of 1:5,000. Alexa Fluor-488–conjugated donkey anti-goat IgG antiserum was obtained from Molecular Probes, Inc., and used at a dilution of 1:600.

Reverse Transcription-PCR Analysis

Total RNA was extracted with Trizol reagent (Invitrogen). The first-strand cDNA was synthesized with 250 ng of random hexamers (Invitrogen), 5 μg of RNA, and SuperScript II reverse transcriptase (Invitrogen) in the presence of 40 units of RNase Inhibitor (Promega). Amplification was carried out with Taq polymerase (Promega) and specific ssDNA primers for BLM (5′-gaagatgctcaggaaagtgac-3′ and 5′-gcagtatgtttattctgatctttc-3′) and 18S rRNA (QuantumRNA 18S Internal Standards, Ambion) as an internal control for 25 PCR cycles (denaturation at 95°C for 40 s, annealing at 56°C for 30 s, extension at 72°C for 40 s).

SCE Assays and Chromosomal Aberration Analysis

Cells were transfected as indicated. After 24 to 40 h of siRNA transfection, cells were transferred to slides and cultured in the presence of 10 μmol/L 5-bromodeoxyuridine (Sigma) at 37°C in an atmosphere containing 5% CO2. After 40 h, colchicine (Sigma) was added to a final concentration of 0.1 μg/mL and the cells were incubated for 1 h. Cells were incubated in hypotonic solution [1:6 (v/v) FCS-distilled water] and fixed with a 3:1 (v/v) mixture of methanol-acetic acid. Cells were then stained by incubation with 10 μg/mL Hoechst 33258 (Sigma) in distilled water for 20 min, rinsed with 2× SSC (Euromedex), exposed to UV light at 365 nm at a distance of 10 cm for 105 min, rinsed in distilled water, stained by incubation with 2% Giemsa solution (VWR) for 16 min, rinsed in distilled water, dried, and mounted. Chromosomes were observed with a Leica DMRB microscope at ×100 magnification. Metaphases were captured with a SONY DXC 930 P camera and SCEs or chromosomal aberrations were analyzed.

Immunofluorescence Microscopy

Immunofluorescence staining was done as previously described (25). Nuclear DNA was detected by mounting slides in Prolong Gold antifade reagent containing DAPI (Invitrogen). Cell images were acquired with a three-dimensional deconvolution imaging system consisting of a Leica DM RXA microscope equipped with a piezoelectric translator (PIFOC, PI) placed at the base of a 63× PlanApo 1.4 numerical aperture objective and a 5-MHz Micromax 1300Y interline charge-coupled device camera (Roper Instruments). Stacks of conventional fluorescence images were collected automatically at a Z-distance of 0.2 μm (Metamorph software, Molecular Devices). Images were generated from 5 to 12 stacked images with Image J (NIH) software.

Live-Cell Imaging

Cells were imaged in DMEM, in six-well plates, in a humidified atmosphere containing 5% CO2 at 37°C. Images were acquired over 0.25 s, every 4 min for 72 h, using one of the following systems: two-dimensional, consisting of a 20×/0.4 or 40×/0.5 objective on a Leica DMIRBE equipped with a 100-W mercury lamp, GFP-band-pass filter set (Chroma), and a Photometrics CoolSNAP cf camera, or four-/five-dimensional, consisting of a 20×/0.4 or 40×/0.5 objective on a Leica DMIRBE equipped with a Polychrome IV (Till Photonics) monochromator, GFP-band-pass filter set (Chroma), and a Photometrics CoolSNAP HQ2 camera. Single-fluorescence and single-phase contrast modes were used to follow cells in six fields of view per well. Image J (26) was used for phenotype analysis.

Statistical Methods

Significance was assessed with Student's t tests or χ2 tests. For all tests, P < 0.05 was considered statistically significant.

BLM Depletion in HeLa Cells Reproduces the Cellular Features Associated with Bloom's Syndrome

We developed a new cellular model for Bloom's syndrome, with control cells containing endogenous BLM protein. Because 81% of BLM mutations introduce a premature termination codon resulting in an absence of BLM protein (27), we stably downregulated BLM in HeLa cells with a shRNA specific for BLM (shBLM). BLM downregulation was highly efficient as the resulting HeLashBLM cells had no detectable BLM protein (Fig. 1A) and had significantly lower BLM mRNA levels (Fig. 1B) than control cells (HeLaV). However, the only slightly higher frequency of SCE indicated that BLM repression in HeLashBLM cells was not complete (Fig. 1C). We increased SCE levels further by transiently downregulating BLM gene expression in HelashBLM cells with a pool of siRNAs directed against sequences other than that targeted by shBLM (HeLash-siBLM). This resulted in lower BLM mRNA levels (Fig. 1B) and significantly higher SCE levels (0.94 SCEs per chromosome) than in control cells (HeLaV-sictrl; 0.22 SCEs per chromosome; Fig. 1C). Moreover, both HelashBLM cells and HeLash-siBLM cells formed colonies less efficiently and grew more slowly than control cells (Fig. 1D and data not shown). Excess SCE in BLM-deficient cells has been shown to be recombination dependent (2, 3). We found that RAD51 downregulation decreased the frequency of SCEs to wild-type levels in BLM-depleted cells (Fig. 1E), showing a strict dependence on RAD51-mediated HR of the increase in SCE in our new Bloom's syndrome cellular model. These characteristics are consistent with the Bloom's syndrome cell phenotype.

FIGURE 1.

BLM-depleted HeLa cells display the cellular features of Bloom's syndrome. A and B, HeLaV and HeLashBLM cells were left untransfected (−) or were transfected with the indicated siRNAs. A, BLM protein levels were assessed by immunoblotting. B, BLM gene expression was monitored by reverse transcription-PCR analysis. The 18S probe was used as loading control. C, number of SCEs per chromosome in cells transfected as in A and B. Between 2,245 and 2,642 chromosomes from four independent experiments were analyzed per condition; bars, SEM. D, HeLaV or HeLashBLM cells treated as in A and B were plated in triplicate at three different dilutions (200, 400, and 600 cells). Columns, mean of three independent experiments; bars, SEM. Significance was assessed with Student's t tests. E, HeLaV or HeLashBLM cells were cotransfected with the indicated siRNA. BLM and RAD51 protein levels were assessed by immunoblotting (left). Number of SCEs per chromosome in cells transfected as in A. Between 680 and 2,181 chromosomes from two independent experiments were analyzed for each set of condition (right); bars, SEM.

FIGURE 1.

BLM-depleted HeLa cells display the cellular features of Bloom's syndrome. A and B, HeLaV and HeLashBLM cells were left untransfected (−) or were transfected with the indicated siRNAs. A, BLM protein levels were assessed by immunoblotting. B, BLM gene expression was monitored by reverse transcription-PCR analysis. The 18S probe was used as loading control. C, number of SCEs per chromosome in cells transfected as in A and B. Between 2,245 and 2,642 chromosomes from four independent experiments were analyzed per condition; bars, SEM. D, HeLaV or HeLashBLM cells treated as in A and B were plated in triplicate at three different dilutions (200, 400, and 600 cells). Columns, mean of three independent experiments; bars, SEM. Significance was assessed with Student's t tests. E, HeLaV or HeLashBLM cells were cotransfected with the indicated siRNA. BLM and RAD51 protein levels were assessed by immunoblotting (left). Number of SCEs per chromosome in cells transfected as in A. Between 680 and 2,181 chromosomes from two independent experiments were analyzed for each set of condition (right); bars, SEM.

Close modal

RAD51 Downregulation in BLM-Depleted Cells Is Associated with a Complete Inhibition of Clonal Growth and Acts Upstream from BLM to Prevent Chromosomal Instability

RAD51 is an essential gene in vertebrate cells (20, 28, 29). Because siRNA-mediated RAD51 downregulation was not lethal in our experimental conditions but was efficient enough to inhibit completely the increase in SCEs in BLM-depleted cells, we further analyzed the effects of RAD51 downregulation in control cells and BLM-depleted cells. We found that RAD51 downregulation significantly decreased the colony-forming efficiency of HeLa control cells and fully abolished the clonal cell growth ability of BLM-depleted cells (Fig. 2A). Thus, unlike cells in which only one of the two proteins is downregulated, cells in which both BLM and RAD51 are downregulated cannot grow in clonal conditions. The downregulation of both BLM and RAD51 therefore had either synergistic or additive effects on clonal growth.

FIGURE 2.

RAD51 downregulation in BLM-depleted cells is associated with a complete inhibition of clonal growth and the induction of chromosomal aberrations. A, cells transfected as in Fig. 1A were plated in triplicate at three dilutions (200, 400, and 600 cells). Columns, mean of four independent experiments; bars, SEM. Significance was assessed with Student's t test. B, the HeLaV and HeLashBLM cells analyzed in Fig. 1E were used for the analysis of chromosomal aberrations. Between 100 and 165 metaphases were analyzed for each set of conditions. Bars, SEM. Significance was assessed with Student's t test. C, typical examples of chromosomal aberrations observed in cells with downregulated RAD51 and/or BLM. Bar, 10 μm. Radials (a), chromosomal constrictions (b), and gaps (c) are indicated (arrows).

FIGURE 2.

RAD51 downregulation in BLM-depleted cells is associated with a complete inhibition of clonal growth and the induction of chromosomal aberrations. A, cells transfected as in Fig. 1A were plated in triplicate at three dilutions (200, 400, and 600 cells). Columns, mean of four independent experiments; bars, SEM. Significance was assessed with Student's t test. B, the HeLaV and HeLashBLM cells analyzed in Fig. 1E were used for the analysis of chromosomal aberrations. Between 100 and 165 metaphases were analyzed for each set of conditions. Bars, SEM. Significance was assessed with Student's t test. C, typical examples of chromosomal aberrations observed in cells with downregulated RAD51 and/or BLM. Bar, 10 μm. Radials (a), chromosomal constrictions (b), and gaps (c) are indicated (arrows).

Close modal

At the chromosomal level, we found that RAD51 downregulation increased the number of cells presenting chromosomal aberrations in both cell lines (Fig. 2B and C; Supplementary Table S1). RAD51 downregulation induced chromosomal instability more frequently (36.6% of cells) than did BLM downregulation (7.2% of cells; Fig. 2B and C). The downregulation of both RAD51 and BLM yielded a frequency of metaphases showing chromosomal aberrations (21.4%) not significantly different from that observed in the presence of RAD51 downregulation alone (36.6%, P = 0.18), indicating that RAD51 acts upstream from BLM to prevent chromosomal instability. With the exception of the radial chromosomes present in our analysis, the chromosomal aberrations we detected in cells with downregulated RAD51 were of a nature and frequency similar to those reported by Schwartz et al. (20), suggesting that at least some of the gaps, breaks, and constrictions occurred at fragile sites.

RAD51 Downregulation Is Associated with an Increase in the Frequency of BLM-Positive Anaphase Bridges

BLM localizes to UFBs and to DAPI-positive anaphase bridges, some of which are associated with fragile sites (17, 19). It has recently been proposed that BLM is required for the resolution of fragile site–associated UFBs and anaphase bridges (19). Because RAD51 downregulation leads to a significant increase in fragile-site expression (20), we investigated its effects on both UFBs and anaphase bridge frequency. The percentage of anaphase cells containing UFBs and the mean number of UFBs per anaphase cell were similar in control cells and cells in which RAD51 was downregulated (Fig. 3A and B). By contrast, the percentage of anaphase cells containing DAPI-positive anaphase bridges was significantly higher in RAD51-downregulated cells than in control cells (Fig. 3C). In these DAPI-positive anaphase bridge cells, RAD51 downregulation was associated with a strong and significant increase in the frequency of BLM-positive anaphase bridges, indicating preferential resolution of the structure of the anaphase bridges formed in response to RAD51 downregulation by BLM (Fig. 3D and E). These data show that RAD51 is required to prevent anaphase bridge formation and that BLM is recruited to anaphase bridges in cells in which RAD51 is downregulated. These results strongly suggest that BLM is required for the suppression or resolution of the anaphase bridges induced by RAD51 downregulation and, thus, that BLM acts downstream from RAD51 to rescue the anaphase bridges resulting from RAD51 deficiency.

FIGURE 3.

RAD51 downregulation is associated with an increase in the frequency of BLM-positive anaphase bridges. HeLaV cells were transfected as in Fig. 2A and fixed as previously described (25). UFBs and DNA bridges were detected by BLM staining. A, percentage of anaphase control cells (n = 321 anaphase cells) and cells displaying RAD51 downregulation (n = 444 anaphase cells) containing UFBs (left). Columns, mean number of UFBs per UFB-positive anaphase cell in control cells (n = 45 UFB-positive anaphase cells) and in cells in which RAD51 was downregulated (n = 72 UFB-positive anaphase cells; right). B, representative examples of Z-projections of multiple stacking images showing the localization of BLM to DAPI-negative DNA bridges, thus revealing UFB-containing anaphase cells. Bar, 10 μm. C, percentage of control cells (n = 321 anaphase cells) and cells displaying RAD51 downregulation (n = 444 anaphase cells) containing DAPI-positive anaphase bridges. Significance was determined by carrying out χ2 test. D, percentage of DAPI-positive anaphase bridges with BLM in control cells (n = 191 DAPI-positive anaphase bridges) and in cells displaying RAD51 downregulation (n = 391 DAPI-positive anaphase bridges). Significance was determined by carrying out χ2 test. E, representative examples of Z-projections of multiple stacking images showing the localization of BLM to DAPI-positive DNA bridges. Bar, 10 μm.

FIGURE 3.

RAD51 downregulation is associated with an increase in the frequency of BLM-positive anaphase bridges. HeLaV cells were transfected as in Fig. 2A and fixed as previously described (25). UFBs and DNA bridges were detected by BLM staining. A, percentage of anaphase control cells (n = 321 anaphase cells) and cells displaying RAD51 downregulation (n = 444 anaphase cells) containing UFBs (left). Columns, mean number of UFBs per UFB-positive anaphase cell in control cells (n = 45 UFB-positive anaphase cells) and in cells in which RAD51 was downregulated (n = 72 UFB-positive anaphase cells; right). B, representative examples of Z-projections of multiple stacking images showing the localization of BLM to DAPI-negative DNA bridges, thus revealing UFB-containing anaphase cells. Bar, 10 μm. C, percentage of control cells (n = 321 anaphase cells) and cells displaying RAD51 downregulation (n = 444 anaphase cells) containing DAPI-positive anaphase bridges. Significance was determined by carrying out χ2 test. D, percentage of DAPI-positive anaphase bridges with BLM in control cells (n = 191 DAPI-positive anaphase bridges) and in cells displaying RAD51 downregulation (n = 391 DAPI-positive anaphase bridges). Significance was determined by carrying out χ2 test. E, representative examples of Z-projections of multiple stacking images showing the localization of BLM to DAPI-positive DNA bridges. Bar, 10 μm.

Close modal

BLM and RAD51 Have Synergistic Effects on Cell Survival

We investigated the possible functional consequences of the increase in BLM localization to anaphase bridges in cells with RAD51 downregulation, using a HeLa cell line stably expressing an H2B-GFP (H2BG) transgene (22) for time-lapse live microscopy analysis. We confirmed that the transient downregulation of BLM and/or RAD51 in HeLa H2BG cells (H2BG-siBLM or H2BG-siRAD51 cells) reproduced the features observed in HeLashBLM cells (Fig. 4A; Supplementary Fig. S1).

FIGURE 4.

BLM and RAD51 act synergistically in cell survival. A, H2BG cells were transfected as in Fig. 2A. BLM and RAD51 protein levels were assessed by immunoblotting. B, H2BG cells were transfected as in Fig. 2A. After 24 h, cells were replated and left untreated (−) or were treated with 2 mmol/L hydroxyurea for 16 h (+). Cells were washed, released into a drug-free medium, and filmed for 72 h. More than 400 mitosis from three independent experiments were analyzed for each set of conditions (a total of 4,803 mitosis analyzed). Cells displaying anaphase bridges were counted. Significance was assessed with Student's t tests. Bars, SEM.

FIGURE 4.

BLM and RAD51 act synergistically in cell survival. A, H2BG cells were transfected as in Fig. 2A. BLM and RAD51 protein levels were assessed by immunoblotting. B, H2BG cells were transfected as in Fig. 2A. After 24 h, cells were replated and left untreated (−) or were treated with 2 mmol/L hydroxyurea for 16 h (+). Cells were washed, released into a drug-free medium, and filmed for 72 h. More than 400 mitosis from three independent experiments were analyzed for each set of conditions (a total of 4,803 mitosis analyzed). Cells displaying anaphase bridges were counted. Significance was assessed with Student's t tests. Bars, SEM.

Close modal

We filmed the cells over 72 hours and found a significantly higher frequency of lagging chromatin and anaphase bridges in H2BG-siBLM cells (3% and 6.08%, respectively) than in control cells (H2BG-sictrl cells; 1.96% and 2.04%, respectively), as expected (refs. 17, 30; Supplementary Fig. S1C and D). We confirmed that RAD51 downregulation (Fig. 4A) resulted in a significant increase in the frequency of anaphase bridge cells (Fig. 4B). This increase was much more marked in H2BG-siBLM cells than in HeLash-siBLM cells due to the lower frequency of anaphase bridge–containing cells among cells overexpressing H2B-GFP (data not shown). RAD51 downregulation also resulted in a significantly higher frequency of anaphase bridge cells (34.73%) than did BLM downregulation (6.08%; Fig. 4B). The downregulation of both BLM and RAD51 resulted in a frequency of anaphase bridge cells (28.34%) similar to that for RAD51 downregulation alone (34.73%, P = 0.097). Thus, BLM is not required to prevent the formation of anaphase bridges in cells in which RAD51 is downregulated.

We investigated whether the formation of anaphase bridges in cells with RAD51 downregulation could result from structures generated in response to disturbances of replication; we did this by carrying out the same experiments in parallel in H2BG cells treated with hydroxyurea, an inhibitor of ribonucleotide reductase commonly used to induce replicational stress. Hydroxyurea blockade was confirmed by fluorescence-activated cell sorting analysis (data not shown). Hydroxyurea-induced replication stress significantly increased the frequency of anaphase bridges in H2BG-sictrl cells (5.67% versus 2.04%) and H2BG-siBLM cells (20.63% versus 6.08%), but not in cells with RAD51 downregulation (Fig. 4B). In conditions of RAD51 downregulation, the frequency of H2BG-sictrl cells (34.73%) and H2BG-siBLM cells (28.34%) presenting anaphase bridges was not significantly modified by hydroxyurea treatment (34.95% and 33.57%, respectively; Fig. 4B).

These results suggest that the formation of anaphase bridges induced by RAD51 downregulation is not due to abnormal structures resulting from the disturbance of replication, being instead due to constitutive DNA structures that are difficult to replicate in normal conditions, such as fragile sites.

We then followed 853 cells with anaphase bridges (420 untreated and 433 hydroxyurea treated) from a total of 4,803 mitotic cells (2,574 untreated and 2,229 hydroxyurea treated) from three independent experiments until division was completed. All the anaphase bridges in control cells and in cells in which BLM was downregulated, either untreated or treated with hydroxyurea, were resolved during telophase or cytokinesis and were associated with the formation of micronuclei, which are known to be present at high frequency in Bloom's syndrome cells (31). By contrast, the resolution of some of the anaphase bridges resulted in cell death in both untreated and hydroxyurea-treated cells in which RAD51 was downregulated, either alone or together with BLM (Supplementary Movies S1 and S2). Indeed, 7% of anaphase bridge cells downregulated for RAD51 and 18% of anaphase bridge cells downregulated for both BLM and RAD51 died just after dividing. After hydroxyurea treatment, 26% of anaphase bridge cells downregulated for RAD51 and 43% of anaphase bridge cells downregulated for both BLM and RAD51 died after dividing, indicating that anaphase bridge cells downregulated for RAD51 display an increase in sensitivity to hydroxyurea (Table 1). Thus, for anaphase bridge cells, death was significantly more frequent among cells downregulated for both BLM and RAD51 than in cells with RAD51 downregulation only, with no cell death recorded for BLM downregulation alone. These results show that BLM downregulation increases the cell death levels associated with RAD51 downregulation, indicating that the inhibition of clonal growth we observed in the cells dowregulated for both BLM and RAD51 results from a synergistic effect, rather than an additive effect, of the depletion of both proteins (Fig. 2A).

Table 1.

Percentage of anaphase bridge cells dying after division

Transfected siRNAsPercent of anaphase bridge cells dying after division
UntreatedAfter hydroxyurea
Ctrl-ctrl 
BLM-ctrl 
Ctrl-RAD51 26 
BLM-RAD51 18 43 
Transfected siRNAsPercent of anaphase bridge cells dying after division
UntreatedAfter hydroxyurea
Ctrl-ctrl 
BLM-ctrl 
Ctrl-RAD51 26 
BLM-RAD51 18 43 

NOTE: Anaphase cells scored in Fig. 4B were followed until division was completed and the frequency of cell death was determined.

Thus, the downregulation of BLM and RAD51 together clearly had a synergistic effect on cell survival, with BLM limiting the lethality associated with RAD51 downregulation by resolving the anaphase bridges resulting from the RAD51 defect.

We studied a new cellular model for Bloom's syndrome consisting of HeLa cells constitutively expressing a shRNA specific for BLM and transiently transfected with a pool of siRNAs directed against sequences other than that targeted by shBLM. These cells display the growth defect, cytogenetic features (high levels of SCE), and mitotic abnormalities (anaphase bridges and lagging chromosomes) typical of Bloom's syndrome cells.

We found that SCE formation in response to BLM depletion was strictly dependent on RAD51, showing an epistatic interaction between the two genes. In mammalian cells, RAD51 involvement has been reported in the formation of induced, but not spontaneous, SCEs (32), suggesting that the SCEs in Bloom's syndrome cells are induced. Thus, the constitutively high levels of RAD51-mediated SCEs in Bloom's syndrome cells are probably induced by the decrease in fork velocity in these cells, which may render them susceptible to DNA breaks (33, 34). RAD51 would then be recruited to the ssDNA leading to a recombination-dependent replication pathway restoring broken forks (15).

Unexpectedly, we found that the increase in chromosomal aberrations or fragile sites induced by RAD51 downregulation was associated with an increase in the frequency of anaphase bridges and of BLM-positive anaphase bridges, but not of BLM-associated UFBs. These results show that UFBs are not derived from RAD51-dependent recombination intermediates and strongly suggest that BLM is required for the suppression or resolution of the anaphase bridges induced by RAD51 downregulation. This conclusion was further supported by our results showing that cells downregulated for both BLM and RAD51 were unable to grow in clonal conditions, whereas cells downregulated for either of the proteins alone could grow clonally. In nonclonal growth conditions, the frequency of cell death in cells downregulated for both BLM and RAD51 was more than 2.5 times that in cells downregulated for RAD51 alone, no cell death being detected among cells downregulated for BLM alone. These results clearly show a synergistic effect of the downregulation of both BLM and RAD51 on cell survival. They also indicate that BLM downregulation increases the cell death associated with RAD51 downregulation, indicating that BLM limits the lethality associated with RAD51 deficiency. An absence of RAD51 is lethal for mammalian cells (29). However, siRNA-mediated RAD51 downregulation was not lethal in our experimental conditions, indicating that RAD51 was still expressed, but at very low levels. This situation mimics the situation in cells underexpressing RAD51 and may have particularly important implications for these cells. Thus, our findings could be of particular importance in cells underexpressing RAD51. Several lines of evidence suggest potential associations between variants of RAD51 family genes and specific forms of cancer (35). Moreover, the RAD51 gene is located at chromosome position 15q15.1 (36), a region displaying a loss of heterozygosity in many cancers (37, 38). Changes in RAD51 gene expression have also been observed in both primary tumors and cancer cell lines (39). BLM activity is therefore essential in such situations to counteract RAD51 underexpression.

We propose a model, based on our findings, explaining the genetic interactions between BLM and RAD51 (Fig. 5). The replication fork slows down, tending to stall or collapse at difficult-to-replicate DNA sequences, such as fragile sites, leading to double-strand break formation. In the presence of RAD51 (RAD51+), the broken forks may be reactivated or repaired by the RAD51-mediated HR pathway, leading to the formation of a Holliday junction that is resolved by the branch migration activity of BLM (Fig. 5A). In the absence of BLM, the RAD51-mediated Holliday junction may be resolved in part by crossing over between chromatids, leading to SCE (Fig. 5A′). The epistatic interaction between RAD51 and BLM observed here is consistent with this pathway, in which BLM acts downstream from RAD51. In RAD51-deficient cells (RAD51), the broken forks remain unrepaired, leading to chromosome breaks and anaphase bridge formation. BLM, together with other proteins such as topoisomerase IIIα, Rmi1, and Rmi2, localizes to some of the anaphase bridges formed in response to RAD51 downregulation and resolves them, thereby limiting cell death (Fig. 5B). The absence of BLM results in a higher frequency of cell death in anaphase bridge–containing cells, accounting for the synergistic effect of the downregulation of both RAD51 and BLM (Fig. 5B′).

FIGURE 5.

A model of the respective roles of RAD51 and BLM at difficult-to-replicate DNA sequences. Black and gray lines, parental template DNA and newly synthesized DNA, respectively. See the text for details.

FIGURE 5.

A model of the respective roles of RAD51 and BLM at difficult-to-replicate DNA sequences. Black and gray lines, parental template DNA and newly synthesized DNA, respectively. See the text for details.

Close modal

In conclusion, our data indicate that BLM acts downstream from RAD51, either to resolve RAD51-mediated Holliday junctions or to rescue anaphase bridges resulting from RAD51 deficiency. BLM and RAD51 must therefore be involved in complex functional interactions because they act epistatically in the maintenance of chromosome integrity and synergistically in cell survival.

No potential conflicts of interest were disclosed.

We thank Giuseppe Baldacci for invaluable stimulating discussions and advice and for constant support, Sarah Lambert for critical reading of the manuscript and helpful advice, and all the members of UMR 2027 for helpful discussions.

Grant Support: The Institut Curie (PIC), the CNRS, INCa grant PL003, and the ARC, and fellowships from the Cancéropôle/Région Ile-de-France, the ARC, and the Ligue contre le Cancer (Comité de l'Essonne; K. Lahkim Bennani-Belhaj), the Ministère de la Recherche (S. Rouzeau), and the CNRS (P. Chabosseau).

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.

1
Ellis
NA
,
Groden
J
,
Ye
TZ
, et al
. 
The Bloom's syndrome gene product is homologous to RecQ helicases
.
Cell
1995
;
83
:
655
66
.
2
Otsuki
M
,
Seki
M
,
Inoue
E
, et al
. 
Functional interactions between BLM and XRCC3 in the cell
.
J Cell Biol
2007
;
179
:
53
63
.
3
Srivastava
V
,
Modi
P
,
Tripathi
V
,
Mudgal
R
,
De
S
,
Sengupta
S
. 
BLM helicase stimulates the ATPase and chromatin-remodeling activities of RAD54
.
J Cell Sci
2009
;
122
:
3093
103
.
4
Sonoda
E
,
Sasaki
MS
,
Morrison
C
,
Yamaguchi-Iwai
Y
,
Takata
M
,
Takeda
S
. 
Sister chromatid exchanges are mediated by homologous recombination in vertebrate cells
.
Mol Cell Biol
1999
;
19
:
5166
9
.
5
van Brabant
AJ
,
Ye
T
,
Sanz
M
,
German
IJ
,
Ellis
NA
,
Holloman
WK
. 
Binding and melting of D-loops by the Bloom syndrome helicase
.
Biochemistry
2000
;
39
:
14617
25
.
6
Singh
TR
,
Ali
AM
,
Busygina
V
, et al
. 
BLAP18/RMI2, a novel OB-fold-containing protein, is an essential component of the Bloom helicase-double Holliday junction dissolvasome
.
Genes Dev
2008
;
22
:
2856
68
.
7
Machwe
A
,
Xiao
L
,
Groden
J
,
Orren
DK
. 
The Werner and Bloom syndrome proteins catalyze regression of a model replication fork
.
Biochemistry
2006
;
45
:
13939
46
.
8
Xu
D
,
Guo
R
,
Sobeck
A
, et al
. 
RMI, a new OB-fold complex essential for Bloom syndrome protein to maintain genome stability
.
Genes Dev
2008
;
22
:
2843
55
.
9
Wu
LHI
. 
The Bloom's syndrome helicase suppresses crossing over during homologous recombination
.
Nature
2003
;
426
:
870
4
.
10
Ralf
C
,
Hickson
ID
,
Wu
L
. 
The Bloom's syndrome helicase can promote the regression of a model replication fork
.
J Biol Chem
2006
;
281
:
22839
46
.
11
Bugreev
DV
,
Yu
X
,
Egelman
EH
,
Mazin
AV
. 
Novel pro- and anti-recombination activities of the Bloom's syndrome helicase
.
Genes Dev
2007
;
21
:
3085
94
.
12
Tripathi
V
,
Nagarjuna
T
,
Sengupta
S
. 
BLM helicase-dependent and -independent roles of 53BP1 during replication stress-mediated homologous recombination
.
J Cell Biol
2007
;
178
:
9
14
.
13
Ouyang
KJ
,
Woo
LL
,
Zhu
J
,
Huo
D
,
Matunis
MJ
,
Ellis
NA
. 
SUMO modification regulates BLM and RAD51 interaction at damaged replication forks
.
PLoS Biol
2009
;
7
:
e1000252
.
14
Bachrati
CZHI
. 
RecQ helicases: guardian angels of the DNA replication fork
.
Chromosoma
2008
;
117
:
219
33
.
15
Amor-Gueret
M
. 
Bloom syndrome, genomic instability and cancer: the SOS-like hypothesis
.
Cancer Lett
2006
;
236
:
1
12
.
16
Wu
L
. 
Role of the BLM helicase in replication fork management
.
DNA Repair (Amst)
2007
;
6
:
936
44
.
17
Chan
KL
,
North
PS
,
Hickson
ID
. 
BLM is required for faithful chromosome segregation and its localization defines a class of ultrafine anaphase bridges
.
EMBO J
2007
;
26
:
3397
409
.
18
Bayart
E
,
Dutertre
S
,
Jaulin
C
,
Guo
RB
,
Xi
XG
,
Amor-Gueret
M
. 
The Bloom syndrome helicase is a substrate of the mitotic Cdc2 kinase
.
Cell Cycle
2006
;
5
:
1681
6
.
19
Chan
KL
,
Palmai-Pallag
T
,
Ying
S
,
Hickson
ID
. 
Replication stress induces sister-chromatid bridging at fragile site loci in mitosis
.
Nat Cell Biol
2009
.
20
Schwartz
M
,
Zlotorynski
E
,
Goldberg
M
, et al
. 
Homologous recombination and nonhomologous end-joining repair pathways regulate fragile site stability
.
Genes Dev
2005
;
19
:
2715
26
.
21
Dutertre
S
,
Ababou
M
,
Onclercq
R
, et al
. 
Cell cycle regulation of the endogenous wild type Bloom's syndrome DNA helicase
.
Oncogene
2000
;
19
:
2731
8
.
22
Kanda
T
,
Sullivan
KF
,
Wahl
GM
. 
Histone-GFP fusion protein enables sensitive analysis of chromosome dynamics in living mammalian cells
.
Curr Biol
1998
;
8
:
377
85
.
23
Mergui
X
,
Leteurtre
F
,
Lipinski
M
,
Bénard
J
,
Amor-Guéret
M
. 
Two distinctly altered cellular responses to DNA double-strand breaks in human neuroblastoma
.
Biochimie
2008
;
90
:
1656
66
.
24
Ababou
M
,
Dumaire
V
,
Lecluse
Y
,
Amor-Gueret
M
. 
Bloom's syndrome protein response to ultraviolet-C radiation and hydroxyurea-mediated DNA synthesis inhibition
.
Oncogene
2002
;
21
:
2079
88
.
25
Wu
L
,
Davies
SL
,
North
PS
, et al
. 
The Bloom's syndrome gene product interacts with topoisomerase III
.
J Biol Chem
2000
;
275
:
9636
44
.
26
Abramoff
MD
,
Magelhaes
PJ
,
Ram
SJ
. 
Image processing with Image
.
J Biophotonics International
2004
;
11
:
36
42
.
27
German
J
,
Sanz
MM
,
Ciocci
S
,
Ye
TZ
,
Ellis
NA
. 
Syndrome-causing mutations of the BLM gene in persons in the Bloom's Syndrome Registry
.
Hum Mutat
2007
;
28
:
743
53
.
28
Sonoda
E
,
Sasaki
MS
,
Buerstedde
JM
, et al
. 
Rad51-deficient vertebrate cells accumulate chromosomal breaks prior to cell death
.
EMBO J
1998
;
17
:
598
608
.
29
Lim
DS
,
Hasty
P
. 
A mutation in mouse rad51 results in an early embryonic lethal that is suppressed by a mutation in p53
.
Mol Cell Biol
1996
;
16
:
7133
43
.
30
German
J
. 
Bloom's syndrome. I. Genetical and clinical observations in the first twenty-seven patients
.
Am J Hum Genet
1969
;
21
:
196
227
.
31
Rosin
MP
,
German
J
. 
Evidence for chromosome instability in vivo in Bloom syndrome: increased numbers of micronuclei in exfoliated cells
.
Hum Genet
1985
;
71
:
187
91
.
32
Lambert
S
,
Lopez
BS
. 
Role of RAD51 in sister-chromatid exchanges in mammalian cells
.
Oncogene
2001
;
20
:
6627
31
.
33
Rao
VA
,
Conti
C
,
Guirouilh-Barbat
J
, et al
. 
Endogenous γ-H2AX-ATM-Chk2 checkpoint activation in Bloom's syndrome helicase deficient cells is related to DNA replication arrested forks
.
Mol Cancer Res
2007
;
5
:
713
24
.
34
Davies
SL
,
North
PS
,
Hickson
ID
. 
Role for BLM in replication-fork restart and suppression of origin firing after replicative stress
.
Nat Struct Mol Biol
2007
;
14
:
677
9
.
35
Thacker
J
. 
The RAD51 gene family, genetic instability and cancer
.
Cancer Lett
2005
;
219
:
125
35
.
36
Takahashi
E
,
Matsuda
Y
,
Hori
T
, et al
. 
Chromosome mapping of the human (RECA) and mouse (Reca) homologs of the yeast RAD51 and Escherichia coli recA genes to human (15q15.1) and mouse (2F1) chromosomes by direct R-banding fluorescence in situ hybridization
.
Genomics
1994
;
19
:
376
8
.
37
Wick
W
,
Petersen
I
,
Schmutzler
RK
, et al
. 
Evidence for a novel tumor suppressor gene on chromosome 15 associated with progression to a metastatic stage in breast cancer
.
Oncogene
1996
;
12
:
973
8
.
38
Gonzalez
R
,
Silva
JM
,
Dominguez
G
, et al
. 
Detection of loss of heterozygosity at RAD51, RAD52, RAD54 and BRCA1 and BRCA2 loci in breast cancer: pathological correlations
.
Br J Cancer
1999
;
81
:
503
9
.
39
Yoshikawa
K
,
Ogawa
T
,
Baer
R
, et al
. 
Abnormal expression of BRCA1 and BRCA1-interactive DNA-repair proteins in breast carcinomas
.
Int J Cancer
2000
;
88
:
28
36
.