SUMO modification regulates BLM and RAD51 interaction at damaged replication forks

Abstract

The gene mutated in Bloom's syndrome, BLM, is important in the repair of damaged replication forks, and it has both pro- and anti-recombinogenic roles in homologous recombination (HR). At damaged forks, BLM interacts with RAD51 recombinase, the essential enzyme in HR that catalyzes homology-dependent strand invasion. We have previously shown that defects in BLM modification by the small ubiquitin-related modifier (SUMO) cause increased gamma-H2AX foci. Because the increased gamma-H2AX could result from defective repair of spontaneous DNA damage, we hypothesized that SUMO modification regulates BLM's function in HR repair at damaged forks. To test this hypothesis, we treated cells that stably expressed a normal BLM (BLM+) or a SUMO-mutant BLM (SM-BLM) with hydroxyurea (HU) and examined the effects of stalled replication forks on RAD51 and its DNA repair functions. HU treatment generated excess gamma-H2AX in SM-BLM compared to BLM+ cells, consistent with a defect in replication-fork repair. SM-BLM cells accumulated increased numbers of DNA breaks and were hypersensitive to DNA damage. Importantly, HU treatment failed to induce sister-chromatid exchanges in SM-BLM cells compared to BLM+ cells, indicating a specific defect in HR repair and suggesting that RAD51 function could be compromised. Consistent with this hypothesis, RAD51 localization to HU-induced repair foci was impaired in SM-BLM cells. These data suggested that RAD51 might interact noncovalently with SUMO. We found that in vitro RAD51 interacts noncovalently with SUMO and that it interacts more efficiently with SUMO-modified BLM compared to unmodified BLM. These data suggest that SUMOylation controls the switch between BLM's pro- and anti-recombinogenic roles in HR. In the absence of BLM SUMOylation, BLM perturbs RAD51 localization at damaged replication forks and inhibits fork repair by HR. Conversely, BLM SUMOylation relieves its inhibitory effects on HR, and it promotes RAD51 function.

Figure 1. HU treatment induces more γ-H2AX in SM-BLM compared to BLM+ cells.

(A) Graphical representation of the average numbers of γ-H2AX foci in HU-treated and untreated BLM+ and SM-BLM cells. BLM+ and SM-BLM cells were untreated (NT) or treated with 0.5 mM HU for 24 h (HU) and then processed for immunofluorescence. Data presented are the average of three experiments in three BLM+ and three SM-BLM clones. Approximately 35 cells were collected in each experiment. Error bars represent standard deviations of the combined data. p-Values were calculated using mixed effects linear models as described in Materials and Methods. *p = 0.02, NT SM-BLM versus NT BLM+; **p = 0.02, HU-treated SM-BLM versus HU-treated BLM+. (B) Excess γ-H2AX-bright cells accumulate in HU-treated SM-BLM cells. Immunofluorescence images of a representative γ-H2AX-bright cell stained with antibodies to γ-H2AX (left panel) and with DAPI (right panel), showing GFP-BLM fluorescence (center panel). SM-BLM localizes to many discrete foci in γ-H2AX-bright cells. Bar indicates 10 µm. (C) Graphical representation of percentages of γ-H2AX-bright cells in BLM+ and SM-BLM cells after no treatment (NT) or treatment (HU) with 0.5 mM HU for 24 h. Error bars represent standard deviations of the combined data. *p = 0.06, NT SM-BLM versus NT BLM+; **p<10⁻³, HU-treated SM-BLM versus HU-treated BLM+. (D) Western blot analysis of γ-H2AX levels in HU-treated and untreated cells. Hsc70 is a control for protein loading. (E) Flow cytometry analysis of γ-H2AX-stained BLM+ and SM-BLM cells. Inset numerical values denote median fluorescence intensities in arbitrary units.

Figure 2. SUMO-mutant BLM cells accumulate excess DSBs at damaged replication forks.

(A) Pulsed-field gel showing the analysis of untreated BLM+ and SM-BLM cells (NT) and cells treated with HU for 24 h (0 time) followed by release into normal medium for 12 and 24 h. DSBs were visualized by DNA fragments that migrate into the gel, whereas intact DNA was retained in the well. (B) Quantification of DSBs after treatment with HU and release into normal medium. The bars represent the numbers of DSBs relative to untreated BLM+ cells in two independent experiments with two clones of each type, and the error bars represent the standard deviations of the combined data. (C) Quantification of accumulation of DSBs after treatment with different concentrations of CPT for 3 h. Data points represent the numbers of DSBs relative to untreated BLM+ or SM-BLM cells from a minimum of two experiments with two clones of each type. Error bars represent standard deviations of the combined data. (D) Quantification of HU-induced micronuclei. Cells were untreated or treated with 0.5 mM HU for 24 h, followed by incubation with cytochalasin-B for 28 h in the absence of HU. The numbers of micronuclei were counted in binucleated cells. A minimum of 500 binucleated cells were assessed under each condition. Data presented are the average of three experiments in three BLM+ and three SM-BLM clones. The bars represent mean numbers of micronuclei, and error bars represent standard deviations of the combined data. *p = 0.6, NT SM-BLM versus NT BLM+; **p<10⁻³, HU-treated SM-BLM versus HU-treated BLM+.

Figure 3. SM-BLM cells are hypersensitive to DNA damage encountered during S phase.

Cells were untreated (NT) or treated with 0.5 mM HU for 24 h and then untreated or treated with 50 µM etoposide for 24 h. Cell viability was measured using the Guava ViaCount reagent, and cell death was calculated as the percentage of dead cells divided by the total number of cells. The bars represent the mean percentage of dead cells in each condition in three BLM+ clones and three SM-BLM clones from a minimum of three independent experiments performed in triplicate. Error bars represent standard deviations of the combined data. See text for p-values of relevant comparisons.

Figure 4. Cells that express SUMO-mutant BLM do not exhibit increased HU-induced SCEs.

BrdU was incorporated for one cell-division cycle prior to HU treatment. Cells were treated with 0.5 mM HU for 24 h and then released into BrdU-containing medium for another 20 h. Metaphases were collected in colcemid. The bars represent the mean numbers of SCEs in two BLM+ and two SM-BLM clones from a minimum of three independent experiments. Error bars represent standard deviations of the combined data. See text for p-values of relevant comparisons.

Figure 5. RAD51 has a localization defect in HU-treated SM-BLM cells.

(A) Immunofluorescence images of BLM+ and SM-BLM cells treated with 0.5 mM HU for 24 h or not treated (NT) and stained for γ-H2AX and RAD51. Bars indicate 10 µm. (B) Graphical representation of mean numbers of RAD51 and BLM foci in HU-treated and untreated BLM+ and SM-BLM cells. *p = 0.09, NT SM-BLM versus NT BLM+; **p = 0.02, HU-treated SM-BLM versus HU-treated BLM+. #p = 0.01, NT SM-BLM versus NT BLM+; ##p = 0.01, HU-treated SM-BLM versus HU-treated BLM+. (C) Graphical representation of mean numbers of colocalized RAD51-γ-H2AX and colocalized BLM-γ-H2AX foci in HU-treated and untreated BLM+ and SM-BLM cells. *p<10⁻³, NT SM-BLM versus NT BLM+; **p<10⁻³, HU-treated SM-BLM versus HU-treated BLM+; #p<10⁻³, NT SM-BLM versus NT BLM+; ##p = 0.02, HU-treated SM-BLM versus HU-treated BLM+. Bars in (B) and (C) represent the average foci in three independent experiments of three BLM+ and three SM-BLM clones. Error bars represent the standard deviations of the combined data. (D) Graphical representation of means numbers of RAD51 foci in untreated cells or cells treated with 10 mM HU for 1 h. RAD51 foci were counted in cells that stained positively with PCNA antibodies. Bars represent the average foci in two independent experiments of two BLM+ and two SM-BLM clones. Error bars represent the standard deviations of the combined data.

Figure 6. RAD51 contains a SUMO binding site.

(A) Biotinylated SUMO-1 and SUMO-2 were bound to streptavidin beads and incubated with purified recombinant, human RAD51; the complexes were pulled down and analyzed by Western blot with anti-RAD51. As a control, unbiotinylated SUMO-1 and SUMO-2 were incubated with beads and incubated with RAD51. (B) A purified, recombinant N-terminal fragment of BLM, consisting of amino acids 1–431, was modified with SUMO-2 in vitro. Equal amounts of SUMOylated or unSUMOylated BLM were incubated with RAD51 bound to streptavidin beads, and complexes were pulled down and analyzed by Western blot with anti-BLM. As a control, SUMOylated or unSUMOylated BLM was incubated with unconjugated streptavidin beads. (C) Ratios of SUMO-2–modified BLM to unmodified BLM in input fraction and fractions bound to control beads and beads containing RAD51. Experiment was performed three times. *p = 0.03, RAD51-coated beads versus control beads.

Figure 7. BLM SUMOylation regulates BLM and RAD51 function in HR-mediated repair of damaged forks.

The model depicts a replication fork that has either stalled or broken. At stalled forks, unSUMOylated BLM inhibits access of RAD51 to the stalled fork. If the fork progresses to a DSB, BLM SUMOylation promotes the recruitment and/or retention of RAD51 to the broken end through noncovalent interactions between SUMOylated BLM and the SUMO binding site on RAD51. S, SUMO.