Summary
Genetic information encoded in chromosomal DNA is challenged by intrinsic and exogenous sources of DNA damage. DNA double-strand breaks (DSBs) are extremely dangerous DNA lesions. RAD51 plays a central role in homologous DSB repair, by facilitating the recombination of damaged DNA with intact DNA in eukaryotes. RAD51 accumulates at sites containing DNA damage to form nuclear foci. However, the mechanism of RAD51 accumulation at sites of DNA damage is still unclear. Post-translational modifications of proteins, such as phosphorylation, acetylation and ubiquitylation play a role in the regulation of protein localization and dynamics. Recently, the covalent binding of small ubiquitin-like modifier (SUMO) proteins to target proteins, termed SUMOylation, at sites containing DNA damage has been shown to play a role in the regulation of the DNA-damage response. Here, we show that the SUMOylation E2 ligase UBC9, and E3 ligases PIAS1 and PIAS4, are required for RAD51 accretion at sites containing DNA damage in human cells. Moreover, we identified a SUMO-interacting motif (SIM) in RAD51, which is necessary for accumulation of RAD51 at sites of DNA damage. These findings suggest that the SUMO–SIM system plays an important role in DNA repair, through the regulation of RAD51 dynamics.
Introduction
Double-strand breaks (DSBs) in chromosomal DNA, caused by both exogenous and endogenous factors such as ionizing radiation and stalled replication forks, are a serious hazard to the survival of cells. Homologous recombination (HR) is an evolutionally conserved mechanism for DSB repair. RAD51 is one of the essential proteins for HR in eukaryotes. Biochemical experiments have shown that RAD51 acts as a recombinase in HR, by forming helical filaments on single-stranded DNA (Wyman and Kanaar, 2006; Atwell et al., 2012). In S-phase cells, RAD51 forms higher-order nuclear structures, called RAD51 nuclear foci, that increase in number after the induction of DNA damage by ionizing irradiation or genotoxic drugs (Haaf et al., 1995; Tashiro et al., 1996; Tashiro et al., 2000). Moreover, RAD51 becomes concentrated at sites containing DSBs induced by UVA-laser microirradiation (Tashiro et al., 2000; Bekker-Jensen et al., 2006). These findings indicated that RAD51 accumulates at sites containing DSBs, where the HR pathway is activated. However, the mechanism that controls the dynamics of RAD51 localization upon DSB formation is unknown.
Post-translational modifications often regulate protein localization and dynamics. Among such protein modifications, the covalent binding of small ubiquitin-like modifier (SUMO) proteins to target proteins has been shown to modulate protein functions, by altering protein–protein interactions as well as the subcellular localization of proteins. Mammalian cells possess three ubiquitous SUMO isoforms: SUMO-1, SUMO-2 and SUMO-3; of these, the latter two are highly conserved and considered to be functionally redundant. SUMOylation is mediated by a cascade of enzymes consisting of the activating enzyme (E1), conjugating enzyme (E2) and ligating enzyme (E3). In mammalian cells, UBC9 is the only SUMOylation E2 enzyme, and the E3 ligases are considered to be important for proper activity and target selection (Cubeñas-Potts and Matunis, 2013). Several lines of evidence suggest the involvement of the SUMO modification system in DNA repair (Kalocsay et al., 2009; Galanty et al., 2009; Morris et al., 2009; Nagai et al., 2011; Ulrich, 2012; Jackson and Durocher, 2013). SUMOylation of PCNA inhibits PCNA-dependent DNA repair of lesions induced by UV exposure or methylmethane sulfonate treatment in budding yeast (Hoege et al., 2002). SUMOylated PCNA functionally cooperates with Srs2, a helicase that blocks the activity of the HR pathway by disrupting RAD51 nucleoprotein filaments, to prevent the unwanted recombination of replicating chromosomes in yeast (Stelter and Ulrich, 2003). Overexpression of SUMO-1 downregulates the DNA DSB-induced HR pathway in Chinese hamster ovary cells and H1080 human fibrosarcoma cells (Li et al., 2000). SUMOylation of XRCC4, a core factor in DNA repair by non-homologous end joining, is important for cell resistance to ionizing radiation and for V(D)J recombination (Yurchenko et al., 2006). Moreover, recent studies revealed the accumulation of SUMO proteins together with UBC9, as well as the E3 enzymes PIAS1 and PIAS4, at DSB sites (Morris et al., 2009; Galanty et al., 2009; Pinder et al., 2013).
SUMOylation plays a role in the accumulation of repair factors at sites containing DNA damage, called DNA-damage-induced foci (Polo and Jackson, 2011). The E3 enzymes PIAS1 and PIAS4 are involved in the SUMOylation and formation of DNA-damage-induced foci containing p53 binding protein 1 (53BP1) and the BRCA1 tumor suppressor protein, after induction of DNA damage by UV microirradiation or genotoxic agents (Galanty et al., 2009; Morris et al., 2009). The SUMOylation-negative mutation of BLM, a RecQ DNA helicase that is defective in Bloom syndrome patients, affects the formation of DNA-damage-induced BRCA1 and γ-H2AX foci (Eladad et al., 2005; Manthei and Keck, 2013). RPA70, a subunit of the replication protein A complex (RPA), is SUMOylated after the induction of replication stress by camptothecin treatment, which facilitates the formation of DNA-damage-induced RAD51 foci (Dou et al., 2010). A mutation at the SUMOylation site of the MDC1 checkpoint protein impairs the production of DNA-damage-induced RPA foci (Luo et al., 2012). After the induction of DNA damage, the ubiquitin ligases HERC2 and RNF168 are modified with SUMO-1 in a PIAS4-dependent manner, and promote ubiquitylation to promote protein assembly at sites of DNA damage induced by ionizing radiation (Danielsen et al., 2012). RNF4, a mammalian SUMO-targeted ubiquitin E3 ligase, also accumulates at sites containing DNA damage after UV microirradiation, where it regulates protein turnover and exchange, and is required for effective DSB repair (Galanty et al., 2012; Yin et al., 2012).
Several lines of evidence suggest a relationship between the function of RAD51 and the SUMOylation system (Li et al., 2000). Among the SUMO proteins, SUMO-1 was identified as a protein that interacts with RAD51 (Shen et al., 1996b). The interaction of RAD51 with UBC9 suggests the involvement of the SUMOylation system in the regulation of RAD51 function (Shen et al., 1996a; Saitoh et al., 2002). Interestingly, the SUMOylation of BLM regulates its interaction with RAD51 and facilitates the HR-mediated repair of stalled replication forks (Ouyang et al., 2009). Although these findings suggest the possible role of the SUMOylation system in RAD51 accumulation at sites containing DNA damage, the means by which the SUMOylation system regulates RAD51 function are still unclear.
In our studies of the regulation of RAD51 dynamics in response to DNA damage, we examined the involvement of the SUMOylation system in the RAD51 accumulation at DNA lesions. Here, we show that the accumulation of RAD51 at damaged DNA sites induced by laser microirradiation is regulated by the SUMO E2 enzyme UBC9 and the E3 enzymes PIAS1 and PIAS4. These findings suggest the involvement of the SUMOylation system in the accumulation of RAD51 at sites containing DSBs. Moreover, we identified a RAD51 SUMO-interacting (SIM) domain required for the DNA-damage-induced accumulation of RAD51. The DNA-damage-dependent activation of the SUMOylation system could therefore function in the regulation of the dynamics of RAD51 localization through the SUMO-SIM interaction.
Results
UBC9 is required for the accumulation of RAD51 at DNA-damage sites
RAD51 has been reported to interact with UBC9, the SUMOylation E2 enzyme (Fig. 1A; Kovalenko et al., 1996). Notably, we identified endogenous UBC9 as a component of the FLAG-HA-RAD51 interactome, purified from nuclear extracts of HeLa cells after γ-irradiation (Fig. 1B). These results led us to investigate whether UBC9 is involved in the regulation of the DNA-damage-induced relocalization of RAD51. Therefore, we decided to apply the UVA-microirradiation system to examine the dynamics of the localization of EGFP-UBC9. Increased signal intensity of EGFP-UBC9 along the microirradiated lines was observed within 1 minute after UVA-microirradiation (Fig. 1C). This suggests that UBC9 accumulates at sites containing DSBs, in agreement with a previous report (Galanty et al., 2009).
UBC9 is required for the DNA-damage-dependent accumulation of RAD51. (A) Co-immunoprecipitation of Myc-UBC9 and RAD51. Anti-Myc immunoprecipitation was performed using GM0637 cells transiently expressing Myc-UBC9 and/or FLAG-HA-RAD51, and then co-precipitating RAD51 was detected by western blotting with anti-RAD51. (B) The RAD51 complex was immuno-affinity purified from the nuclear soluble fraction of HeLa S3 cells stably expressing FLAG-HA-RAD51, at 30 minutes after ionizing irradiation (12 Gy) treatment. The immunoblotting analysis was performed using an anti-UBC9 antibody. (C) Quantitative analysis of EGFP signal intensity after UVA microirradiation in GM0637 cells, expressing either EGFP-UBC9 (left) or EGFP-SUMO-1 (right). The signal intensity of EGFP in the microirradiated areas was measured. To confirm the DSB-dependent accumulation of UBC9 at microirradiated sites, the cells were pre-incubated in medium containing Hoechst 33342 (+Hoechst) or not (−Hoechst) before UVA-microirradiation. Results are means ± s.e. of three independent experiments. (D) Depletion of UBC9 by siRNA against UBC9. Immunofluorescence staining of GM0637 cells transfected with the UBC9-specific siRNA [siUBC9 (#1)] was performed using an anti-UBC9 antibody. UBC9 and DNA are shown in red and blue, respectively, in the merged images. Scale bars: 20 µm. (E) Immunoblotting analysis of GM0637 cells transfected with siUBC9 (#1), using an anti-UBC9 antibody. (F) Immunofluorescence staining of UBC9-depleted GM0637 cells, at 25 minutes after UVA microirradiation, was performed using anti-RAD51 and anti-γH2AX antibodies. RAD51, γH2AX and DNA (Hoechst 33342) are shown in green, red and blue, respectively, in the merged images. Scale bars: 20 µm. (G) Quantification of RAD51 accumulation at microirradiated regions in GM0637 cells. The mean of three independent experiments is indicated, with the error bars representing the s.d.
UBC9 is required for the DNA-damage-dependent accumulation of RAD51. (A) Co-immunoprecipitation of Myc-UBC9 and RAD51. Anti-Myc immunoprecipitation was performed using GM0637 cells transiently expressing Myc-UBC9 and/or FLAG-HA-RAD51, and then co-precipitating RAD51 was detected by western blotting with anti-RAD51. (B) The RAD51 complex was immuno-affinity purified from the nuclear soluble fraction of HeLa S3 cells stably expressing FLAG-HA-RAD51, at 30 minutes after ionizing irradiation (12 Gy) treatment. The immunoblotting analysis was performed using an anti-UBC9 antibody. (C) Quantitative analysis of EGFP signal intensity after UVA microirradiation in GM0637 cells, expressing either EGFP-UBC9 (left) or EGFP-SUMO-1 (right). The signal intensity of EGFP in the microirradiated areas was measured. To confirm the DSB-dependent accumulation of UBC9 at microirradiated sites, the cells were pre-incubated in medium containing Hoechst 33342 (+Hoechst) or not (−Hoechst) before UVA-microirradiation. Results are means ± s.e. of three independent experiments. (D) Depletion of UBC9 by siRNA against UBC9. Immunofluorescence staining of GM0637 cells transfected with the UBC9-specific siRNA [siUBC9 (#1)] was performed using an anti-UBC9 antibody. UBC9 and DNA are shown in red and blue, respectively, in the merged images. Scale bars: 20 µm. (E) Immunoblotting analysis of GM0637 cells transfected with siUBC9 (#1), using an anti-UBC9 antibody. (F) Immunofluorescence staining of UBC9-depleted GM0637 cells, at 25 minutes after UVA microirradiation, was performed using anti-RAD51 and anti-γH2AX antibodies. RAD51, γH2AX and DNA (Hoechst 33342) are shown in green, red and blue, respectively, in the merged images. Scale bars: 20 µm. (G) Quantification of RAD51 accumulation at microirradiated regions in GM0637 cells. The mean of three independent experiments is indicated, with the error bars representing the s.d.
Because RAD51 accumulates at sites containing DNA damage within 10–20 minutes after UVA-microirradiation (Tashiro et al., 2000), the accumulation of UBC9 at sites of DNA damage occurs before that of RAD51 (Fig. 1C; supplementary material Fig. S1). Therefore, we next investigated the requirement of UBC9 for the accumulation of RAD51 at DNA-damage sites. For this, we used short-interfering RNAs (siRNAs) to deplete UBC9 in the human fibroblast cell line GM0637 and the human osteosarcoma cell line U2OS (Fig. 1D,E; supplementary material Fig. S2A,D). Immunofluorescence staining using an anti-RAD51 antibody revealed that depletion of UBC9 significantly reduced the percentage of cells with increased RAD51 signal intensity at the microirradiated sites (Fig. 1F,G; supplementary material Fig. S2B,E,F). HR is active and RAD51 forms nuclear foci in the S/G2 phase, and depletion of UBC9 does not decrease the percentage of cells in S/G2 phase (Hayashi et al., 2002; Riquelme et al., 2006; Tashiro et al., 1996; Durant and Nickoloff, 2005). Therefore, our results suggest the involvement of UBC9 in the accumulation of RAD51 at sites containing DNA damage. To confirm this possibility, we examined the RAD51 focus formation in UBC9-depleted U2OS cells after treatment with ionizing irradiation, and found that RAD51 focus formation was significantly disturbed by depletion of UBC9 (supplementary material Fig. S2C). These results suggest that UBC9 is required for RAD51 accumulation at sites containing DNA damage.
SUMO-1 is required for RAD51 accumulation at sites containing DNA damage
RAD51 interacts preferentially with SUMO-1, as compared with SUMO-2 (supplementary material Fig. S3) (Shen et al., 1996b). Importantly, SUMO-1 accumulated at sites containing DNA damage induced by UVA microirradiation (supplementary material Fig. S1). Therefore, to elucidate the role of SUMO modifications in the regulation of RAD51 function, we examined RAD51 accretion at sites containing DSBs in SUMO-1-depleted cells after UVA microirradiation (Fig. 2A–C). The percentage of cells with endogenous RAD51 accumulation at UVA-microirradiated sites was significantly lower in SUMO-1-depleted cells compared with the control (Fig. 2B,C). The involvement of SUMO-1 in the accumulation of RAD51 at damaged DNA sites was also supported by the finding that SUMO-1 depletion significantly reduced the percentage of cells with RAD51 foci after γ-irradiation (Fig. 2D). These findings indicate that SUMO-1 is required for RAD51 accumulation at sites with DNA damage.
SUMO-1 is involved in DNA-damage-dependent accumulation of RAD51. (A) Depletion of SUMO-1 by pSIREN-shSUMO-1 (arrows). Cells transfected with pSIREN-shSUMO-1 are DsRed-positive. Endogenous SUMO-1 was detected by immunofluorescence staining, using an anti-SUMO-1 antibody. Scale bars: 20 µm. (B) Damage-dependent accumulation of RAD51 (arrows) was visualized by immunofluorescence staining, using an anti-RAD51 antibody, in control and SUMO-1-depleted cells at 30 minutes after UVA-microirradiation. RAD51, DsRed and DNA (Hoechst 33342) are shown in green, red and blue, respectively, in the merged images. Scale bars: 20 µm. (C) Percentage of cells showing RAD51 accumulation at microirradiated regions. Results are the total of two independent experiments. n = 99 for vector, n = 66 for shSUMO-1 (#1) and n = 78 for shSUMO-1 (#2), respectively. (D) Percentage of RAD51-foci-positive cells expressing either shSUMO-1 (#1) or shSUMO-1 (#2) at 2 hours after γ-irradiation (2 Gy). The mean of three independent experiments is indicated, with error bars representing the s.d.
SUMO-1 is involved in DNA-damage-dependent accumulation of RAD51. (A) Depletion of SUMO-1 by pSIREN-shSUMO-1 (arrows). Cells transfected with pSIREN-shSUMO-1 are DsRed-positive. Endogenous SUMO-1 was detected by immunofluorescence staining, using an anti-SUMO-1 antibody. Scale bars: 20 µm. (B) Damage-dependent accumulation of RAD51 (arrows) was visualized by immunofluorescence staining, using an anti-RAD51 antibody, in control and SUMO-1-depleted cells at 30 minutes after UVA-microirradiation. RAD51, DsRed and DNA (Hoechst 33342) are shown in green, red and blue, respectively, in the merged images. Scale bars: 20 µm. (C) Percentage of cells showing RAD51 accumulation at microirradiated regions. Results are the total of two independent experiments. n = 99 for vector, n = 66 for shSUMO-1 (#1) and n = 78 for shSUMO-1 (#2), respectively. (D) Percentage of RAD51-foci-positive cells expressing either shSUMO-1 (#1) or shSUMO-1 (#2) at 2 hours after γ-irradiation (2 Gy). The mean of three independent experiments is indicated, with error bars representing the s.d.
Requirement of PIAS1 and PIAS4 for RAD51 accumulation at DNA-damage sites
The SUMOylation E3 ligases PIAS1 and PIAS4 function in the DNA-damage response (Galanty et al., 2009). Therefore, we next examined whether PIAS1 or PIAS4 are involved in the accumulation of RAD51 at DNA-damage sites. For this purpose, we examined the effect of depletion of PIAS1 or PIAS4 by siRNA on the accumulation of RAD51 at microirradiated sites (Fig. 3). Immunofluorescence staining of UVA-microirradiated cells revealed that the accretion of RAD51 at the microirradiated sites was significantly impaired by the depletion of PIAS4 (Fig. 3C,D). In PIAS1-depleted cells, the accumulation of RAD51 at microirradiated sites was also decreased, but not as much as in PIAS4-depleted cells (Fig. 3D). Because the percentage of cells in S/G2 is not decreased by the depletion of PIAS1 or PIAS4 (Galanty et al., 2009), these findings suggest that PIAS1 and PIAS4 are important for the regulation of RAD51 dynamics in the DNA-damage response. Taken together, our results suggest that activation of the SUMOylation system at DNA-damage sites is required for the localization of RAD51.
PIAS1 and PIAS4, the E3 enzymes of SUMOylation, are important for DNA-damage-dependent accumulation of RAD51. (A) Depletion of PIAS1 by siRNA in GM0637 cells was confirmed by immunoblotting, using an anti-PIAS1 antibody. (B) Depletion of PIAS4 by siRNA was confirmed with anti-FLAG immunoblotting analysis of GM0637 cells expressing FLAG-PIAS4, transfected with either control or PIAS4 siRNAs. (C) Immunofluorescence staining of PIAS1- or PIAS4-depleted GM0637 cells at 30 minutes after UVA microirradiation, using anti-RAD51 and anti-γH2AX antibodies. RAD51, γH2AX and DNA (Hoechst 33342) are shown in green, red and blue, respectively, in the merged images. Scale bars: 20 µm. (D) Percentage of cells with RAD51 accumulation at microirradiated regions in PIAS1- or PIAS4-depleted GM0637 cells. The mean of three independent experiments is indicated, with error bars representing the s.d.
PIAS1 and PIAS4, the E3 enzymes of SUMOylation, are important for DNA-damage-dependent accumulation of RAD51. (A) Depletion of PIAS1 by siRNA in GM0637 cells was confirmed by immunoblotting, using an anti-PIAS1 antibody. (B) Depletion of PIAS4 by siRNA was confirmed with anti-FLAG immunoblotting analysis of GM0637 cells expressing FLAG-PIAS4, transfected with either control or PIAS4 siRNAs. (C) Immunofluorescence staining of PIAS1- or PIAS4-depleted GM0637 cells at 30 minutes after UVA microirradiation, using anti-RAD51 and anti-γH2AX antibodies. RAD51, γH2AX and DNA (Hoechst 33342) are shown in green, red and blue, respectively, in the merged images. Scale bars: 20 µm. (D) Percentage of cells with RAD51 accumulation at microirradiated regions in PIAS1- or PIAS4-depleted GM0637 cells. The mean of three independent experiments is indicated, with error bars representing the s.d.
RAD51 interacts with SUMO through its SIM domain
Our findings indicate the involvement of SUMOylation in the accumulation of RAD51 at sites of DNA damage. Although RAD51 interacted with SUMO-1 in a yeast two-hybrid analysis, RAD51 lacks putative SUMOylation consensus sequences (Fig. 4B) (Shen et al., 1996b). Therefore, the interaction between RAD51 and SUMO has been suggested to be non-covalent (Ouyang et al., 2009). The SUMO interacting motif (SIM; V/I-A-V/I-V/I) has been identified as the domain required for non-covalent binding with SUMO (Song et al., 2004; Shen et al., 2006; Lin et al., 2006). We found this consensus in RAD51 at amino acid positions 261–264 (VAVV) in the C-terminal region, and determined that it is conserved from yeast to human (Fig. 4A). This led us to study the effect of the SIM in RAD51 on its interaction with SUMO-1. First, we constructed RAD51 mutants in which the putative SIM was mutagenized, by replacing the large nonpolar amino acids (VAVV) with small nonpolar ones (AAAS) or substituting V264 with lysine (V264K), respectively, which has been demonstrated to cause SIM dysfunction (Song et al., 2004; Shen et al., 2006; Lin et al., 2006). To examine the requirement of the SIM of RAD51 for the interaction with SUMO-1, we performed a yeast two-hybrid analysis. Neither RAD51-AAAS nor RAD51-V264K interacted with SUMO-1 (Fig. 4B). These findings suggest that the RAD51 SIM domain is required for the interaction with SUMO-1.
RAD51 interacts with SUMO-1 through its SIM. (A) The SUMO-interacting motif (SIM) is evolutionally conserved in eukaryotic RAD51 homologues. Amino acid residues matching the consensus sequences of SIM (V/I-A-V/I-V/I) are shown in red. (B) Two-hybrid analysis of RAD51 and SUMO-1. The growth of two independent transformants on SD +His (left) and SD –His (middle) plates was examined for each combination of vectors indicated (right).
RAD51 interacts with SUMO-1 through its SIM. (A) The SUMO-interacting motif (SIM) is evolutionally conserved in eukaryotic RAD51 homologues. Amino acid residues matching the consensus sequences of SIM (V/I-A-V/I-V/I) are shown in red. (B) Two-hybrid analysis of RAD51 and SUMO-1. The growth of two independent transformants on SD +His (left) and SD –His (middle) plates was examined for each combination of vectors indicated (right).
The RAD51 SIM domain is required for the regulation of RAD51 dynamics after induction of DNA damage
To address whether the SIM is required for DNA-damage-dependent accumulation of RAD51, we microirradiated cells expressing EGFP-tagged RAD51-WT, RAD51-AAAS or RAD51-V264K. The accumulation of RAD51-WT at sites of DNA damage was observed in more than 40% of the microirradiated cells (18 out of 38 microirradiated cells, supplementary material Fig. S4). By contrast, RAD51-AAAS and RAD51-V264K failed to accumulate at the microirradiated sites (supplementary material Fig. S4). This suggested that SIM is required for the accumulation of RAD51 at DNA damage. Although RAD51 is localized to both the nucleus and cytoplasm (Mladenov et al., 2009), the EGFP-tagged RAD51-AAAS and RAD51-V264K proteins showed cytoplasmic-dominant distributions, compared with RAD51-WT in GM0637 cells (supplementary material Fig. S4A). Thus, the two mutants might fail to accumulate at DNA-damage sites because of their insufficient amounts in the nucleus, rather than their impaired interaction with SUMO. To exclude this possibility, we inserted a nuclear localization signal (NLS) into the EGFP-tagged RAD51 mutants. After confirming their localization was now mostly nuclear, we performed UVA-microirradiation analysis (Fig. 5). The EGFP-NLS-tagged RAD51-WT accumulated at microirradiated sites, but the RAD51-AAAS and RAD51-V264K mutants both failed to do so (Fig. 5). Taken together, these findings support the proposal that the RAD51 SIM plays an important role in RAD51 accumulation at sites containing DNA damage.
The SIM domain of RAD51 is required for the DNA-damage-dependent accumulation of RAD51. (A) EGFP fluorescence signals in GM0637 cells transfected with pEGFP-NLS, pEGFP-NLS-RAD51 (WT), pEGFP-NLS-RAD51-AAAS (AAAS) or pEGFP-NLS-RAD51-V264K (V264K) expression vectors were examined at 1 hour after UVA microirradiation. Immunofluorescence staining of cells with an anti-γH2AX antibody was performed to visualize the microirradiated sites. EGFP, γH2AX and DNA (Hoechst 33342) are shown in green, red and blue, respectively, in the merged images. Arrows indicate the accumulation of EGFP-NLS-RAD51 at the microirradiated sites. Scale bars: 20 µm. (B) Percentage of cells expressing either pEGFP-NLS (n = 31), pEGFP-NLS-RAD51 (n = 33), pEGFP-NLS-RAD51-AAAS (N = 31) or pEGFP-NLS-RAD51-V264K (n = 37), showing the increased signal intensity of EGFP at the microirradiated regions.
The SIM domain of RAD51 is required for the DNA-damage-dependent accumulation of RAD51. (A) EGFP fluorescence signals in GM0637 cells transfected with pEGFP-NLS, pEGFP-NLS-RAD51 (WT), pEGFP-NLS-RAD51-AAAS (AAAS) or pEGFP-NLS-RAD51-V264K (V264K) expression vectors were examined at 1 hour after UVA microirradiation. Immunofluorescence staining of cells with an anti-γH2AX antibody was performed to visualize the microirradiated sites. EGFP, γH2AX and DNA (Hoechst 33342) are shown in green, red and blue, respectively, in the merged images. Arrows indicate the accumulation of EGFP-NLS-RAD51 at the microirradiated sites. Scale bars: 20 µm. (B) Percentage of cells expressing either pEGFP-NLS (n = 31), pEGFP-NLS-RAD51 (n = 33), pEGFP-NLS-RAD51-AAAS (N = 31) or pEGFP-NLS-RAD51-V264K (n = 37), showing the increased signal intensity of EGFP at the microirradiated regions.
The RAD51 SIM is required for the recombinational repair of DSBs
Having established that the RAD51 SIM plays a role in the regulation of RAD51, we asked whether the RAD51 SIM-SUMO interaction functionally affects DSB repair. To this end, we examined the effect of the expression of RAD51-WT or RAD51-V264K on HR levels in U2OS cells, using a DR-GFP assay. In this assay, a single DSB is introduced into the chromosomally integrated DR-GFP substrate by the I-SceI endonuclease, and the repair of the DSB by HR generates cells expressing functional GFP (Pierce and Jasin, 2005; Sakamoto et al., 2007; Hosoya et al., 2012). The transfection of siRNA against RAD51 decreased the percentage of GFP-positive cells by around 50% (supplementary material Fig. S5). As expected, the exogenous expression of RAD51-WT significantly restored the HR activity repressed by depletion of endogenous RAD51 (supplementary material Fig. S5). By contrast, the expression of the RAD51-V264K protein in U2OS-DRGFP cells failed to do so (supplementary material Fig. S5). To exclude the possibility that the failure to restore HR activity was due to the low nuclear amounts of the RAD51 mutants, we performed the DR-GFP assay using NLS-tagged RAD51 WT, V264K and AAAS mutants. As a result, the restoration of the HR activity by RAD51 WT, but not by the RAD51 mutants, was confirmed (Fig. 6). This finding supports the proposal that the SUMO–SIM interaction, involving the RAD51 SIM domain, is required for the regulation of HR activity in cells. Taken together, our findings suggest that the in situ activation of the SUMOylation system around DSBs facilitates the accumulation of RAD51 through its SIM domain for HR activity.
The SIM domain of RAD51 is required for HR. (A) HR activity of RAD51-depleted cells (siRAD51), expressing NLS (vector), NLS-RAD51-WT, NLS-RAD51-V2645K or NLS-RAD51-AAAS, was analyzed using the DR-GFP assay. Columns represent the mean of the proportions of GFP-positive cells from four independent experiments. Error bars represent s.d. (B) Expression of NLS-RAD51-WT, NLS-RAD51-V2645K or NLS-RAD51-AAAS in RAD51-depleted cells (siRAD51) was confirmed by an immunoblot analysis using anti-RAD51 antibodies.
The SIM domain of RAD51 is required for HR. (A) HR activity of RAD51-depleted cells (siRAD51), expressing NLS (vector), NLS-RAD51-WT, NLS-RAD51-V2645K or NLS-RAD51-AAAS, was analyzed using the DR-GFP assay. Columns represent the mean of the proportions of GFP-positive cells from four independent experiments. Error bars represent s.d. (B) Expression of NLS-RAD51-WT, NLS-RAD51-V2645K or NLS-RAD51-AAAS in RAD51-depleted cells (siRAD51) was confirmed by an immunoblot analysis using anti-RAD51 antibodies.
Discussion
We have shown that SUMO-1, the E2 SUMO conjugating enzyme UBC9, and the E3 ligases PIAS1 and PIAS4 are involved in the accumulation of human RAD51 at DSB-containing sites. We also identified SIM, a SUMO-interacting motif in RAD51, which is required for RAD51 accretion at damaged DNA sites. These findings strongly suggest that the SUMO–SIM system, involving the interaction of RAD51 with free SUMO-1 or SUMOylated proteins through the SIM, plays an important role in the regulation of RAD51 dynamics for DNA repair in human cells.
We showed that of the SUMOylation E3 enzymes, PIAS4 contributes more strongly to the accumulation of RAD51 at sites containing DSBs, compared with PIAS1 (Fig. 3). We also found that SUMO-1, rather than SUMO-2, interacts with RAD51 (supplementary material Fig. S3). Because SUMOylation with SUMO-1 is facilitated by PIAS4 upon DNA damage, RAD51 accumulation could be mainly dependent on the PIAS4-mediated modification of proteins at sites of DNA damage with SUMO-1 (Galanty et al., 2009). Importantly, both PIAS1 and PIAS4 are involved in DNA-damage-dependent SUMOylation (Galanty et al., 2009). Recently, DNA damage has been shown to trigger a SUMOylation wave, leading to simultaneous multisite modifications of several DNA repair proteins of the HR pathway in yeast (Psakhye and Jentsch, 2012). The SIM domain of RAD51 is conserved from yeast to human (Fig. 4A), it is therefore also possible that in human cells, such multiple SUMOylation reactions of various proteins at DNA-damage sites could play a role in the regulation of HR. The RAD51 function should be strictly regulated within a very narrow range, because both the overexpression and depletion of RAD51 lead to increased genome instability (Richardson et al., 2004). Several lines of evidence have indicated that in S. cerevisiae the SUMOylation of PCNA increases the interaction with Srs2, leading to disruption of RAD51 nucleoprotein filaments at stalled replication forks to prevent unnecessary HR activity (Stelter and Ulrich, 2003). In addition, overexpression of SUMO-1 has been shown to downregulate HR activity in mammalian cells (Li et al., 2000). Therefore, SUMOylation at sites containing DNA damage could play opposite roles in the regulation of HR, by facilitating the HR reaction by the accumulation of repair proteins, and by repressing inappropriate HR activity, especially around replication forks. If the RAD51 dynamics were regulated by only one protein interaction, then it would be difficult to execute such precise control of its association and dissociation from sites of DNA damage. RAD51 also interacts with HR-related proteins, such as BRCA2, for its dynamic regulation (Yu et al., 2003; Liu et al., 2010; Thorslund et al., 2010; Holloman, 2011). The fine-tuning of the accumulation of RAD51 at damaged DNA sites, by direct multiple interactions with various DNA-repair-associated proteins and/or through the SIM-SUMO system, could facilitate the strict regulation of the HR activity for the appropriate repair of DNA damage.
The radiation-induced nuclear focus formation of RAD51 at the sites containing DNA damage requires SUMOylation enzymes. Because focus formation involves the accumulation and retention of proteins, the SUMOylation system could also regulate the retention of RAD51 at DNA-damage sites, for the formation of higher-order nuclear structures. Several lines of evidence suggest the involvement of SUMOylation in the formation of higher-order nuclear architectures. The covalent binding of SUMO to DNA-repair-related proteins, such as RAD52, TIP60, RAP80 and PCNA, is induced upon DNA damage (Hoege et al., 2002; Sacher et al., 2006; Yan et al., 2007; Cheng et al., 2008). The accumulation of UBC9, PIAS1 and PIAS4 could enhance the SUMOylation of DNA-repair proteins around DSBs, which would help to recruit and retain RAD51 around DSBs through its SIM, leading to the formation of damage-induced RAD51 foci. Three recent papers shed light on the function of SIMs in the DNA-damage response (Galanty et al., 2012; Yin et al., 2012; Praefcke et al., 2012). RNF4, a ubiquitin E3 enzyme, contains SIM domains and targets SUMOylated DNA-repair proteins such as MDC1 and RPA, leading to the appropriate turnover of these proteins through proteasome-dependent proteolysis. RPA-SUMO-2/3 is required for the damage-dependent accumulation of RAD51 (Dou et al., 2010). However, when RNF4 is depleted, RPA-SUMO-2/3 persists at sites of DNA damage and inhibits the subsequent accumulation of RAD51 (Galanty et al., 2012). Therefore, RPA-SUMO-2/3 might not be the direct target of the RAD51 SIM for RAD51 accumulation at damaged sites. Further exploration of the targets of RAD51 SIM will provide clues to elucidate the detailed mechanisms of RAD51 regulation.
In this study, we identified a SIM domain in the C-terminal region of RAD51. This region is important for binding of RAD51 to the nuclear matrix (Nagai et al., 2008; Mladenov et al., 2009). The nuclear matrix is a dynamic subnuclear compartment supporting the spatial organization of DNA metabolism (Anachkova et al., 2005; Zaidi et al., 2007; Courbet et al., 2008). However, according to the chromosome territory interchromatin compartment (CT-IC) model, the nuclear architecture is highly organized and based on two principal components: the chromatin compartment and the interchromatin compartment (Cremer and Cremer, 2010). The chromatin compartment is composed of interconnected chromosome territories, which in turn are formed from interconnected megabase-sized chromatin domains (∼1 Mbp) and larger chromatin clusters (Dixon et al., 2012; Markaki et al., 2012). Transcription reportedly occurs on the surfaces of chromatin domains or the perichromatin region (Markaki et al., 2012). Because some RAD51 foci are located in the interior of the interchromatin compartment, damaged DNA could be moved from the inside of the chromatin compartment to the interchromatin compartment or perichromatin region for the HR reaction, as in the case of transcription (Albiez et al., 2006). Although the role of the nuclear matrix in the regulation of DNA repair is still unclear, binding of RAD51 to the nuclear matrix could regulate the dynamics of RAD51 in the interchromatin compartment for the proper HR reaction (Chiolo et al., 2011).
Materials and Methods
Cell culture and ionizing irradiation
GM0637, an SV40-transformed human fibroblast cell line, and the human osteosarcoma U2OS cell line were cultured in Dulbecco's modified Eagle's medium, supplemented with antibiotics (penicillin and streptomycin, Sigma-Aldrich) and 10% fetal calf serum (Tashiro et al., 2000). Ionizing irradiation was performed using a 137Cs source at the indicated dose.
Antibodies
Rabbit anti-RAD51 (Calbiochem PC130, BioAcademia 70-001), mouse anti-γH2AX (Upstate Biotechnology, 05-636), rabbit anti-UBC9 (Becton Dickinson, 610748), rabbit anti-SUMO-1 (ALEXIS, 210-174-R200), rabbit anti-PIAS1 (Cell Signaling, 3550), rabbit anti-PIASy (Abgent, AP1249a), mouse anti-Myc (Roche 11667149001), FITC-conjugated goat anti-rabbit (Tago Immunological, ALI4408), Cy3-conjugated sheep anti-mouse (Jackson ImmunoResearch, 515-165-062) and Cy5-conjugated goat anti-mouse (Jackson ImmunoResearch 115-176-003) antibodies were used in the experiments.
Plasmids
pEGFP-RAD51 was constructed by inserting the PCR-amplified RAD51 cDNA into the BglII site of pEGFP-c1 (Clontech). To construct pEGFP-NLS-RAD51, two oligonucleotides (5′-CCGGAGCCCCTCCTAAAAAAAAACGGAAAGTCGGGA-3′ and 5′-GATCTCCCGACTTTCCGTTTTTTTTTAGGAGGGGCT-3′) were annealed, and the resulting DNA encoding an NLS (APPKKKRKVG) was inserted into pEGFP-RAD51 via the BspEI/BglII sites. For exogenous expression of RAD51 in the HR assay, NLS-RAD51 was inserted into pcDNA3.1/Myc-His(-)C via its NotI/HindIII sites. The SIM mutations were introduced by site-directed mutagenesis. pGAD424-SUMO-1 and pGAD424-SUMO-2 were kindly provided by Dr T. Nishida (Nishida et al., 2000). pGBK-RAD51 was constructed by inserting the RAD51 cDNA into the BamHI/PstI sites of pGBK.
Protein interactions
To examine the interaction between Myc-UBC9 and FLAG-HA-RAD51, GM0637 cells co-transfected with pcDNA3.1-Myc-UBC9 and pcDNA3.1-FLAG-HA-RAD51 were lysed in 50 mM HEPES-KOH (pH 7.4), 100 mM NaCl, 10 mM EDTA, 0.5% NP-40, 1 mM DTT and 1 mM PMSF. The cleared lysate was mixed with the mouse anti-Myc antibody bound to sheep anti-mouse IgG Dynabeads (Invitrogen) at 4°C for 1.5 hours, and then eluted by 1×SDS-PAGE sample buffer. The co-precipitated RAD51 was detected by western blotting. Establishment of the FLAG-HA-RAD51 stable expression cell line and purification of FLAG-HA-RAD51 from the nuclear soluble fraction were performed as described previously (Nakatani and Ogryzko, 2003).
The Matchmaker two-hybrid system (Clontech) was used for the yeast two-hybrid assay. AH109 cells were transformed with two plasmids, one containing the wild-type or mutant RAD51 gene fused with GAL4 binding domain, and the other containing SUMO-1 fused with the GAL4 activating domain. Transformants were streaked on SD/−Trp/−Leu and SD/−Trp/−Leu/−His plates, and grown at 30°C for 2 days.
UVA-microirradiation and live cell imaging
UVA microirradiation and live-cell imaging analysis was performed using a ZEISS LSM510 confocal laser-scanning microscope, with a 63× /1.4 plan-apochromat objective. Cells on round coverslips were transferred to a Chamlide TC live-cell chamber system (Live Cell Instrument), mounted on the microscope stage, and maintained at 37°C. The objective was operated with a heater, as part of the Chamlide TC. For UVA microirradiation, Hoechst 33258 (Sigma) was added to cultures at 2 µg/ml. After 10 minutes, the DMEM was replaced by Leibovitz's L-15 (Gibco) containing 10% FBS and 25 mM HEPES (Gibco). UVA microirradiation was performed as described previously (Ikura et al., 2007). The 364 nm line of the UVA-laser was used to introduce DSBs. Adobe Photoshop was used for presentation of the images.
Immunofluorescence microscopy
Cells were fixed for 10 minutes with PBS containing 4% paraformaldehyde, and then permeabilized with PBS containing 1% SDS and 0.5% Triton X-100 for 10 minutes. The cells were then incubated with antibodies in PBS containing 1% BSA at 37°C for 30 minutes. Nuclei were stained with Hoechst 33342. An Axioplan2 microscope, with AxioCam MRm controlled by Axiovision (Zeiss), was used for visualization. To quantify protein accumulation after microirradiation, at least 20 cells were microirradiated in each experiment and subjected to immunofluorescence microscopy. The number of nuclei showing colocalization of RAD51 with γH2AX at microirradiated sites was then counted.
RNAi
The pSIREN-DNR-DsRed-Express vector (Clontech) was used for SUMO-1 RNAi. The target sequences were 5′-CUGGGAAUGGAGGAAGAAG-3′ (#1) and 5′-CACAUCUCAAGAAACUCAA-3′ (#2). The experiments were performed 3 days after vector transfection. siGENOME SMARTpool (Thermo Scientific Dharmacon M-004910-00) and Silencer Select siRNA (Ambion s14589, 5′-AAACAGAUCCUAUUAGGAA-3′) were purchased and used as siUBC9 (#1) and siUBC9 (#2), respectively. For PIAS1 and PIAS4 RNAi, the ON-TARGETplus SMART pool (Thermo Scientific Dharmacon L-008167-00 and L-006445-00) was used. siGENOME Non-Targeting siRNA Pool #1 (Thermo Scientific Dharmacon D-001206-13) was used as the control RNA. The experiments were performed 2 days after transfection with 5–10 nM siRNA.
Homologous recombination repair assay
The HR repair assay was performed as previously reported (Kobayashi et al., 2010). RAD51-depleted U2OS-DRGFP cells created using siRNA (Ambion s11734, 5′-GGUAGAAUCUAGGUAUGC-3′) were reconstituted with wild-type RAD51, RAD51V264K or RAD51AAAS mutants resistant to siRNA, with or without an NLS. To measure the HR repair of I-SceI-generated DSBs, 50 µg of the I-SceI expression vector (pCBASce) was introduced to U2OS-DRGFP cells by electroporation (GenePulser; Bio-Rad). To determine the efficiency of HR repair, the percentage of GFP-positive cells was quantified by a FACSCantoII cell sorter (Becton Dickinson) at three days after electroporation.
Note added in proof
During the revision of this manuscript, Dr Stefan Jentsch's lab at the Max Planck Institute of Biochemistry reported the identification of Rad51 SIM in yeast (Bergink et al., 2013).
Acknowledgements
We thank Drs Maria Jasin, Junya Kobayashi and Kenshi Komatsu for providing U2OS-DRGFP cells, and T. Nishida for the pGAD424-SUMO-1 and pGAD424-SUMO-2 vectors.
Author contributions
H. Shima, Y.H., T.I., R.K., K.I., H. Saitoh, H.K. and S.T. planned the experimental design and wrote the manuscript. H. Shima, H. Suzuki, J.S., K.K., L.S., A.K., Y.H, T.I. and M.I. performed experiments.
Funding
This work was supported in part by Grants-in-Aid from the Japanese Society for the Promotion of Science (JSPS), and the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.