G9a coordinates with the RPA complex to promote DNA damage repair and cell survival

Significance

The G9a histone methyltransferase primarily regulates the expression of genes associated with cancer development in cancer cells, but it has also been implicated in mediating the DNA damage response. Here, we confirmed a role for G9a in DNA damage repair following double-strand breaks. G9a is recruited to chromatin as a result of casein kinase 2-mediated phosphorylation, where it directly interacts with replication protein A (RPA). G9a binding to RPA modulates RPA and Rad51 foci formation and permits efficient homologous recombination. This molecular mechanism renders cancer cells more resistant to radiation and chemotherapeutics. Our improved understanding of the molecular function of G9a may help with the future design of G9a inhibitors and G9a-based DNA damage agents as cancer therapeutics.

Abstract

Histone methyltransferase G9a has critical roles in promoting cancer-cell growth and gene suppression, but whether it is also associated with the DNA damage response is rarely studied. Here, we report that loss of G9a impairs DNA damage repair and enhances the sensitivity of cancer cells to radiation and chemotherapeutics. In response to DNA double-strand breaks (DSBs), G9a is phosphorylated at serine 211 by casein kinase 2 (CK2) and recruited to chromatin. The chromatin-enriched G9a can then directly interact with replication protein A (RPA) and promote loading of the RPA and Rad51 recombinase to DSBs. This mechanism facilitates homologous recombination (HR) and cell survival. We confirmed the interaction between RPA and G9a to be critical for RPA foci formation and HR upon DNA damage. Collectively, our findings demonstrate a regulatory pathway based on CK2–G9a–RPA that permits HR in cancer cells and provide further rationale for the use of G9a inhibitors as a cancer therapeutic.

The human genome is constantly challenged by spontaneous DNA metabolism and environmental agents that can induce DNA lesions (1, 2). To maintain genomic integrity, cells have evolved a complex protective mechanism known as the DNA damage response (DDR) to sense signals of DNA damage and repair different types of DNA lesion (3). Of all of the distinctive types of DNA lesion, DNA double-strand breaks (DSBs) are the most cytotoxic, and defects in repairing such lesion can result in genomic instability and lead to cancer (1, 4). There are two major pathways to repair DSBs—homologous recombination (HR) and nonhomologous end joining (NHEJ). HR is an error-free process that can occur only during the S and G2/M phases of the cell cycle. Conversely, NHEJ is an error-prone process that can operate throughout the cell cycle (5, 6). A key step in determining whether HR or NHEJ may occur is DNA end resection—a process that produces 3′ single-stranded DNA (ssDNA) tails (7). These newly created tails bind to Rad51 recombinase, which initiates DNA pairing and strand invasion (8). Replication protein A (RPA) is a heterotrimeric protein complex composed of RPA70, RPA32, and RPA14 that also binds to the 3′ ssDNA tails. Upon binding, RPA activates an ATR–Chk1-dependent G2/M DNA damage checkpoint (9–11). The Mre11–Rad50–Nbs1 (MRN) complex, CtIP (RBBP8), and BRCA1 have all been identified as major regulators that initiate DNA end resection (12–14). Conversely, 53BP1 is thought to prevent DNA end resection and thus facilitates NHEJ (15).

The dense packaging of nuclear DNA into chromatin is an obstacle to DNA repair (16, 17), and many studies have emphasized the importance of chromatin regulators, such as histone methylations and their corresponding modifiers, in the DDR. The levels of histone H3K36 dimethylation (H3K36me2) increase following DSBs as a result of either the dissociation of the histone demethylase KDM2A from chromatin or the catalytic activity of the histone methyltransferase Metnase, and recruit either the MRN complex or Ku70 family proteins to DSBs, respectively (18, 19). Suv39h1-mediated histone H3K9 trimethylation (H3K9me3) at DSBs directly binds to and activates Tip60—a histone acetylase that is essential for ataxia-telangiectasia–mutated (ATM) activation (20, 21). In addition, PR-Set7 cooperates with Suv4-20 to promote histone H4K20 dimethylation (H4K20me2) to permit the binding of 53BP1 to DSBs (22, 23). Thus, it is likely that any enzyme that alters local chromatin structure surrounding the damaged DNA sites will influence the capacity of the DDR.

G9a is the primary enzyme for histone H3K9 monomethylation (H3K9me1) and dimethylation (H3K9me2) in euchromatin (24, 25). Global knockout of G9a in mice is embryonic lethal due to a severe growth limitation (26). In addition, the expression levels of G9a are often up-regulated in various human cancers (27). Depletion of G9a in cancer cells inhibits cell growth, leads to chromosome instability, and markedly reduces tumor growth (28–30). Interestingly, signaling of DNA damage in senescent cells induces proteasome-dependent degradation of G9a (31) and application of a G9a inhibitor (UNC0638) in combination with a low dose of the DNA damage agents phleomycin and etoposide is shown to selectively attenuate cancer-cell proliferation (32). These findings lead us to hypothesize that G9a may be involved in the DDR following DNA damage.

In this study, we identified that G9a can regulate HR in response to DSBs. G9a deficiency impairs DNA damage repair and sensitizes cancer cells to DSBs by disrupting RPA and Rad51 foci formation. This action of G9a is dependent on casein kinase 2 (CK2), which phosphorylates G9a at serine 211 to permit its recruitment to chromatin. In addition, we found that direct binding of G9a to RPA70 is indispensable for G9a function in DNA damage repair and that this effect was independent of its methyltransferase activity. Altogether, our data demonstrate that G9a is directly involved in HR following DNA damage in human cancer cells, which will be useful for the design of new cancer therapeutics, particularly for those patients with elevated G9a levels.

Results

G9a Is Required for DNA Damage Repair.

We first investigated whether G9a is required for DNA damage repair in human colorectal carcinoma HCT116 cells by producing a G9a knockout cell line (here referred to as G9a-KO and the parental cells referred to as G9a-WT) using the CRISPR-Cas9 technique. Although knockout of G9a in mice is embryonic lethal (26), G9a-KO HCT116 cells, similar to G9a knockout embryonic stem cells (26), are viable. The G9a knockout efficiency and specificity was confirmed by Western blotting and DNA sequencing (Fig. S1 A and B). Consistent with previous reports (33, 34), G9a knockout cells exhibited more spontaneous DNA damage and reduced DNA replication and colony formation capacity, thus strengthening the idea that G9a is required for cell proliferation (Fig. S1 C–E). We then used a comet assay to directly measure the efficiency of DNA repair in these cells after inducing DNA damage with irradiation (IR) or etoposide (VP16). As shown in Fig. 1 A and B, there was a negligible difference between the comet tails in G9a-KO and G9a-WT cells 1 h after DNA damage treatment; however, significant DNA damage, as indicated by increased comet-tail length and area (35), was detected in the G9a-KO cells 6 h post IR or 18 h after VP16 withdrawal, compared with G9a-WT cells (P < 0.05).

Fig. S1.

G9a is required for DNA damage repair. This figure is related to Fig. 1. (A) Whole-cell lysates were extracted from G9a-WT cells or the two CRISPR clones G9a-KO#1 and G9a-KO#2 and subjected to Western blotting to detect G9a protein levels. (B) Genomic DNA was extracted from G9a-WT, G9a-KO#1, or G9a-KO#2 cells, respectively. Following PCR and purification, the PCR products were subjected to DNA sequencing. The Upper schematic (a) represents the G9a genomic locus and two sgRNA regions. The Lower schematic (b) shows the sequencing results of sgRNA regions. del, deletion; in, insert; nt, nucleotide. (C) G9a-WT or G9a-KO cells were subjected to immunofluorescence using anti–γ-H2AX antibodies and visualized by confocal microscopy. The statistical diagram below indicates the percentage of cells with Ф H2AX antibo γ-H2AX foci. The data represent the means ± SD (n ≥ 100). (D) Statistical chart of G9a-WT or G9a-KO cells to show the cells with BrdU incorporation. Cells were labeled with BrdU for 30 min. The percentage of BrdU-positive cells is indicated in each column. (E) A representative image of colony formation in G9a-WT or G9a-KO cells. (F) G9a-WT or G9a-KO HCT116 cells were treated with or without IR (10 Gy) and recultured under normal conditions for 1, 3, 6, and 12 h. Whole-cell lysates were subjected to Western blotting with the indicated antibodies. “0 hr” indicates cells with IR treatment but with no time for DNA repair. (G) G9a-WT or G9a-KO HCT116 cells were treated with or without VP16 for 1 h. After VP16 withdrawal, cells were recultured under normal conditions for 18 h. Whole-cell lysates were subjected to Western blotting with the indicated antibodies. “0 hr” indicates cells with VP16 treatment but with no time for DNA repair. (H) HCT116 or U2OS cells were treated with or without IR (10 Gy) and recultured under normal conditions for the indicated time interval. Whole-cell lysates were analyzed with an anti–γ-H2AX antibody. “0 hr” indicates cells with IR treatment but with no time for DNA repair. (I) Mouse embryonic fibroblasts (MEFs) were irradiated (10 Gy) and recultured for the indicated time interval. Western blotting was performed as described in H.

Fig. 1.

G9a is required for DNA damage repair. (A) G9a-WT or G9a-KO HCT116 cells were exposed to IR (10 Gy) or treated with VP16 (40 μM for 1 h) and then allowed to recover under normal conditions for the indicated times before analysis by comet assay. A representative image for each condition is shown. (B) Quantification of the tail moments shown in A. The data represent the means ± SD (n ≥ 100) from three independent experiments. (C) Cells that were treated as described in A were fixed and labeled with an anti–γ-H2AX antibody. (D) Quantification of γ-H2AX–positive cells in C. The γ-H2AX–positive cells refer to cells with ≥10 γ-H2AX foci. The data represent the means ± SD (n ≥ 100) from three independent experiments. (E) Colony formation assay for G9a-WT or G9a-KO cells. The survival fraction was calculated, and the data represent the means ± SD from three independent experiments. Student’s t test (two-tailed): G9a-KO versus G9a-WT, P < 0.05.

H2AX is phosphorylated at Ser139 (γ-H2AX) in response to DNA damage, and the levels of γ-H2AX have been widely used as a sensitive marker for DSBs (36). We therefore analyzed γ-H2AX foci formation in G9a-WT or G9a-KO cells by immunofluorescence. Microscopic analyses identified that γ-H2AX foci were formed in both the G9a-WT and G9a-KO cells 1 h post DNA damage, but at 6 h post IR or 18 h after VP16 withdrawal, the majority of γ-H2AX foci returned to basal levels in the G9a-WT cells, whereas the G9a-KO cells still exhibited visible γ-H2AX foci (P < 0.05), suggesting that lack of G9a impaired DNA damage repair (Fig. 1 C and D). We next performed Western blotting to further confirm the dynamic changes in γ-H2AX levels and obtained similar data to those described above (Fig. S1 F and G). Finally, a colony formation assay was used to determine if G9a depletion sensitizes cancer cells to DSBs. As shown in Fig. 1E, the G9a-KO cells exhibited a much lower survival rate following DNA damage treatment compared with the G9a-WT cells (P < 0.05). Altogether, these data indicate that G9a is involved in DNA damage repair and thus affects cell survival.

G9a Is Recruited to Chromatin in Response to DNA Damage.

Previous work has shown that G9a is degraded in response to DNA damage in primary human diploid fibroblasts (31). Here, the total G9a protein levels did not change in HCT116, HeLa, or LoVo cancer cells following DNA damage (Fig. S2A). Interestingly, we found that G9a moved from the soluble portion of the nucleoplasm to nuclear chromatin upon IR or exposure to the chemotherapeutics VP16 or camptothecin (CPT), suggesting that the movement of G9a to chromatin is a universal response to DSBs (Fig. 2 A and B). In addition, its movement in response to VP16 treatment occurred in a dose-dependent and time-dependent manner (Fig. 2 C and D). To examine the dynamic enrichment of G9a at sites of DNA damage, HCT116 and HeLa cells were transfected with full-length G9a tagged with a GFP epitope (GFP-G9a) and then observed under a fluorescence microscope after irradiation with a 365-nm UV laser beam. As shown in Fig. 2 E and F, G9a accumulation at DNA damage stripes reached a plateau ∼180 s after laser microirradiation. In addition, the enzyme-dead GFP-G9a (Y1154F) protein, in which a tyrosine was changed to phenylalanine and the N terminus GFP-G9a (1–630 aa) protein alone was also recruited to DNA damage stripes, suggesting that the N terminus but not its methylation activity is essential for G9a recruitment (Fig. S2B). A cellular assay using the DR-GFP U2OS reporter system was also used to monitor the recruitment of G9a to DSBs. Briefly, DR-GFP U2OS cells are stably incorporated with two separate GFP genes, one of which contains a recognition site for endonuclease I-SceI. A DSB site can be generated by overexpression of I-SceI, and the recruitment of DNA-responsive proteins at DSBs can be detected by confocal microscopy or chromatin immunoprecipitation (ChIP) (37). Although we were unable to observe visible G9a foci at I-SceI–induced DSB sites, possibly because of the low sensitivity of our anti-G9a antibody, G9a did colocalize with γ-H2AX (Fig. S2D). In addition, the ChIP assay indicated that, following overexpression of I-SceI, G9a was recruited to the DSB site (an approximately twofold increase), and the change in G9a cellular localization was similar to that of Rad51—a DNA damage responsive protein that was used as a positive control (Fig. 2G). These data indicate that G9a is recruited to chromatin in response to DNA damage.

Fig. S2.

G9a is recruited to chromatin in response to DNA damage. This figure is related to Fig. 2. (A) HCT116, HeLa, or LoVo cells were exposed to IR, VP16, or CPT. Whole-cell lysates were subjected to Western blotting with the indicated antibodies. (B) HeLa cells were transfected with GFP-G9a (1154YF) or GFP-G9a (1–630 aa) and irradiated with a 365-nm UV laser beam. Images were collected every 20 s after laser microirradiation. The arrows indicate the irradiation path. (C) HeLa cells transfected with GFP-G9a were pretreated with or without BrdU (20 μM, 24 h) and then irradiated with a 365-nm UV laser beam. Images were collected every 20 s after laser microirradiation. The arrows indicate the irradiation path. (D) DR-U2OS cells transfected with I-SceI were fixed and labeled with anti–γ-H2AX or anti-G9a antibodies. A representative image of the cells with I-SceI overexpression is shown.

Fig. 2.

G9a is recruited to chromatin in response to DNA damage. (A) HCT116 cells were exposed to IR (10 Gy) and then recultured under normal conditions for 1, 3, or 6 h. Dt and Chr proteins were then extracted for Western blotting and analyzed using antibodies against G9a and γ-H2AX. β-Actin and Histone H3 were used as the loading controls for nucleoplasm and chromatin proteins, respectively. “0 hr” indicates that the cells exposed to IR but with no time for DNA repair. (B) HCT116 or HeLa cells were treated with or without VP16 (40 μM) or CPT (2 μM) for 1 h. Protein extraction and Western blotting was performed as in A. “CTR” indicates the cells without any treatment. (C) HCT116 cells were treated with VP16 at the indicated doses for 1 h. (D) HCT116 cells were treated with VP16 (40 μM) for the indicated time course. (E) HCT116 or HeLa cells were transfected with GFP-G9a (WT) and irradiated with a 365-nm UV laser beam. Images were collected every 20 s after laser microirradiation. The arrows indicate the irradiation path. (F) Quantification of the dynamic recruitment of GFP-G9a to damaged DNA stripes in E. The recruitment of m-Cherry-RPA70 to laser beam-induced DNA damage sites was used as a positive control. (G) DR-GFP U2OS cells were transfected with I-SceI. At 24 h after transfection, ChIP assays were performed. Rad51 was used as a positive control. The data shown represent the means ± SD from three independent experiments.

CK2 Catalyzes G9a Phosphorylation in Vivo and in Vitro.

We next explored how G9a is recruited to chromatin in response to DSBs by analyzing the G9a posttranslational modifications that occur upon VP16 or CPT treatment. We found that Serine/Threonine (Ser/Thr) phosphorylation of G9a markedly increased after VP16 or CPT treatment, whereas other posttranslational modifications, including tyrosine (Tyr) phosphorylation and lysine (Lys) methylation, remained unchanged (Fig. 3A). In addition, increases in Ser/Thr phosphorylation were abolished by λ-phosphatase (λ-PPase) treatment, thus supporting the idea that G9a is phosphorylated in response to DNA damage (Fig. 3B). To determine whether the increased levels of Ser/Thr phosphorylation were required for G9a recruitment to chromatin, the G9a protein levels in various subcellular compartments were investigated after treatment with VP16, with or without multiple kinase inhibitors. The recruitment of G9a to chromatin was not affected by inhibitors targeting the PIKK family members ATM, ATR, or DNA-PK (Fig. S3 A–C). However, G9a recruitment to chromatin upon VP16 treatment was attenuated by 4,5,6,7-tetrabromobenzotriazole (TBB)—an inhibitor that specifically targets CK2 (Fig. 3C and Fig. S3D). In addition, HCT116 cells were transiently transfected with an siRNA against CK2 followed by VP16 treatment. As shown in Fig. 3D, siRNA-mediated CK2 knockdown also attenuated G9a enrichment on chromatin in response to VP16 treatment. Moreover, the CK2 inhibitor TBB, but not the ATM inhibitor KU55933, impaired the accumulation of GFP-G9a at the DNA damage stripes (Fig. 3E). To investigate if G9a is phosphorylated by CK2 in vivo, the Ser/Thr phosphorylation levels of G9a were studied in HCT116 cells in which FLAG-tagged WT CK2 [FLAG-CK2 (WT)] or FLAG-tagged CK2 mutant [FLAG-CK2 (K68M)] with defective enzymatic activity was overexpressed. The Ser/Thr phosphorylation levels of exogenous G9a were increased in cells transfected with FLAG-CK2 (WT), but not in cells transfected with FLAG-CK2 (K68M) (Fig. 3F). In addition, the Ser/Thr phosphorylation levels of G9a were no longer increased upon VP16 treatment in HCT116 cells transfected with CK2 siRNA (Fig. 3G). Moreover, the in vitro kinase catalytic assay showed that G9a was effectively phosphorylated by FLAG-CK2 (WT) but not by FLAG-CK2 (K68M) (Fig. 3H). We conclude therefore that CK2 is the enzyme responsible for G9a phosphorylation and its changes in subcellular localization in response to DNA damage.

Fig. 3.

CK2 catalyzes G9a phosphorylation in vivo and in vitro. (A) HCT116 cells were transfected with FLAG-G9a and then treated with VP16, CPT, or vehicle. The FLAG-G9a immunoprecipitates were subjected to Western blotting with the indicated antibodies. FLAG-G9a was used as a loading control. (B) FLAG-G9a immunoprecipitates were incubated with or without λ–PPase in vitro and then analyzed by Western blotting. (C) HCT116 cells were pretreated with or without TBB (50 μM, 12 h) before VP16 (40 μM, 1 h) treatment. Western blotting was performed to determine G9a subcellular localization. (D) HCT116 cells were transfected with an siRNA against CK2 for 48 h and then treated with VP16. Again, Western blotting was performed to detect G9a cellular localization. (E) HeLa cells were transfected with GFP-G9a (WT), pretreated with TBB or KU55933 (10 μM, 6 h), and then irradiated with a 365-nm UV laser beam. Images were collected every 20 s after laser microirradiation. The arrows indicate the irradiation path. (F) HCT116 cells were cotransfected with either GFP-G9a plus FLAG-CK2 (WT) or GFP-G9a plus FLAG-CK2 (K68M) for 48 h. GFP-G9a immunoprecipitates were subjected to immunoblotting with the indicated antibodies. GFP-G9a was used as a loading control. (G) HCT116 cells were transfected with an siRNA against CK2 for 24 h followed by transfection with FLAG-G9a for 48 h. FLAG-G9a immunoprecipitates were subjected to immunoblotting with the indicated antibodies. (H) HCT116 cells were transfected with FLAG-CK2 (WT) or FLAG-CK2 (K68M), and the FLAG immunoprecipitates were incubated with bacterially purified G9a (WT) in vitro. Western blotting was performed with the indicated antibodies to detect G9a phosphorylation.

Fig. S3.

PIKK family kinases are not the enzymes responsible for G9a recruitment to chromatin. This figure is related to Fig. 3. (A–C) HCT116 cells were pretreated with or without inhibitors targeting ATM (A), ATR (B), or DNA-PK (C) before being treated with or without VP16. Chromatin-bound proteins were extracted and subjected to Western blotting with the indicated antibodies. Anti-p53S15ph or anti-RPA32 was used as a positive control. (D) HeLa cells were pretreated with or without TBB before VP16 treatment. Dt and Chr proteins were extracted to detect G9a protein levels.

Phosphorylation of G9a at Ser211 Permits Its Recruitment to Chromatin.

To identify the specific CK2 phosphorylation sites on G9a, we incubated recombinant G9a with functional CK2 and then performed mass spectrometry (MS). These MS data indicated that Serine 211 (Ser211) was the site of CK2 phosphorylation (Fig. S4A). To confirm this result, a G9a Ser-211 to alanine 211 mutant [G9a (S211A)] was generated, and an in vitro phosphorylation assay was performed with either recombinant G9a (WT) or G9a (S211A) as substrates. As shown in Fig. 4A, G9a (WT) was phosphorylated by CK2, but the G9a S211A mutation markedly abrogated G9a phosphorylation. In addition, alignment of the sequence surrounding the Ser211 site in different species indicated that the surrounding sequence is conserved in mammals but not in Danio rerio or Drosophila melanogaster (Fig. S4B). We then used coimmunoprecipitation (Co-IP) to study whether Ser211 phosphorylation is required for G9a recruitment to chromatin upon DNA damage by monitoring the interaction between various G9a mutants (WT, S211A, and S211D) and histone H3. As shown in Fig. 4B, the interaction between G9a (S211A) and H3 was markedly decreased compared with both the G9a (WT) and G9a (S211D) mutant, in which a change from serine to aspartic acid mimics phosphorylation. In addition, cells expressing G9a (S211A) exhibited reduced G9a chromatin enrichment, whereas the enrichment of G9a (WT) and G9a (S211D) on chromatin was comparable (Fig. 4C). Moreover, the translocation of G9a from the nucleoplasm to chromatin upon VP16 treatment was attenuated in the presence of the S211A mutation (Fig. 4D). Furthermore, live-cell imaging revealed that the S211A mutation caused reduced recruitment of G9a to sites of DNA damage, whereas the G9a (S211D) mutation enhanced its accumulation at damaged DNA sites (Fig. 4 E and F). Importantly, the methyltransferase activity of G9a remained unchanged between G9a (WT) and G9a (S211A) (Fig. S4C). Taken together, these findings indicate that phosphorylation of G9a at Ser211 promotes its recruitment to chromatin by enhancing its chromatin-binding capacity but not by altering its methyltransferase activity.

Fig. S4.

Phosphorylation of G9a at Ser211 permits its recruitment to chromatin and leads to increased H3K9me2 levels. This figure is related to Fig. 4. (A) CK2 catalyzes G9a at Ser211. Bacterially purified G9a full-length (GST-tag was cut by HRV3C) was catalyzed by FLAG-CK2 in vitro and subsequently separated by SDS/PAGE and stained with CBB. The G9a band was excised from the gel and analyzed by mass spectrometry. (B) The schematic shows the alignment result of the sequence surrounding Ser211 in different species. The gray columns represent the sites conserved in mammals, and the red column shows the Ser211 site in each species. (C) HCT116 cells were transfected with FLAG-G9a (WT) or FLAG-G9a (S211A), and the FLAG immunoprecipitates were incubated with full-length recombinant H3. Western blotting was performed to detect H3K9me2 levels. (D) Acid extraction of histones from HCT116 cells with treatments as described in Fig. 2A. (E) Acid extraction of histones from HCT116 cells with or without VP16 or CPT treatment. (F) G9a-WT or G9a-KO HCT116 cells were treated with or without VP16. Acid-extracted histones were subjected to Western blotting to detect H3K9me2 levels, and whole-cell lysates were analyzed using an anti-G9a antibody.

Fig. 4.

Phosphorylation of G9a at Ser211 permits its recruitment to chromatin. (A) FLAG-CK2 precipitates were incubated with bacterially purified G9a (WT) or G9a (S211A) in vitro, and Western blotting was performed to detect G9a phosphorylation. (B) HCT116 cells were transfected with pcDNA3.1(+) FLAG-G9a (WT), FLAG-G9a (S211A), or FLAG-G9a (S211D). The FLAG-G9a immunoprecipitates were subjected to Western blotting with the indicated antibodies. FLAG-G9a was used as a loading control. “FLAG” indicates FLAG-tagged G9a WT or its mutants. (C) HCT116 cells were transfected with pcDNA3.1(+) FLAG-G9a (WT), FLAG-G9a (S211A), or FLAG-G9a (S211D). Soluble nucleoplasm or chromatin extracts were subjected to Western blotting with the indicated antibodies. (D) HCT116 cells transfected with GFP-G9a (WT) or GFP-G9a (S211A) were treated with or without VP16. The chromatin-bound proteins were extracted and subjected to Western blotting with the indicated antibodies. (E) HeLa cells were transfected with GFP-G9a (WT), GFP-G9a (S211A), or GFP-G9a (S211D) for 24 h and irradiated with a 365-nm UV laser beam. Images were collected every 20 s after laser microirradiation, and the arrows indicate the irradiation path. (F) A quantitative analysis of the dynamic recruitment of GFP-tagged proteins in E. The data represent the means ± SD (n ≥ 20). Student’s t test (two-tailed): S211D versus WT, P < 0.05; S211A versus WT, P < 0.05.

G9a Interacts with RPA in Vivo and in Vitro.

To investigate how chromatin-enriched G9a contributes to DNA damage repair, the possible G9a-interacting proteins were identified in HCT116 cells transfected with FLAG-G9a by MS. We identified several well-known G9a-interacting proteins, including CDYL, WIZ, and GLP (25, 38, 39), as well as the novel binding partners RPA70 and RPA32 (Fig. S5A). To confirm this result, a series of Co-IP assays were performed in HCT116 cells. Consistent with the MS results, endogenous RPA70 and RPA32 were precipitated by exogenous FLAG-G9a (Fig. S5B), and the interaction between them markedly increased upon DNA damage (Fig. 5A). In addition, endogenous G9a and the RPA complex were mutually precipitated by each other under physiological conditions (Fig. 5B and Fig. S5C). Moreover, the endogenous interaction between G9a and RPA decreased in the soluble nucleoplasm fraction and increased in the chromatin-containing fraction upon VP16 treatment (Fig. 5C). To confirm whether RPA70 or RPA32 interacted with G9a, we performed Co-IP assays in HCT116 cells by cotransfecting either GFP-G9a and FLAG-tagged RPA70 (FLAG-RPA70) or GFP-G9a and FLAG-tagged RPA32 (FLAG-RPA32). As shown in Fig. 5 D and E, GFP-G9a and FLAG-RPA70 were mutually precipitated; however, GFP-G9a failed to immunoprecipitate FLAG-RPA32, perhaps implying that G9a has a preferential interaction with RPA70 over RPA32 in vivo. Mutant G9a (S211A) also exhibited reduced binding affinity for RPA70 compared with G9a (WT) and G9a (S211D), suggesting that phosphorylation of G9a at Ser211 facilitates G9a binding to RPA70 (Fig. 5F). To verify whether the interaction between G9a and the RPA complex is direct, a GST pull-down assay was performed. As shown in Fig. 5G and Fig. S5D, G9a directly interacted with RPA70 and RPA32. We then refined the GST pull-down assay to map the mutual interacting domains between G9a and RPA70. As shown in Fig. 5H, full-length G9a, G9a (1–350 aa), and G9a (350–630 aa) were pulled down by RPA70, but not G9a (630–960 aa) or G9a (960–1210 aa), suggesting that G9a interacts with RPA70 via the N terminus [G9a (1–630 aa)]. The interactions between these different G9a fragments and RPA70 were then investigated in vivo. Concordant with the in vitro data, RPA70 exhibited a preferential interaction with G9a (1–630 aa) (Fig. 5I). In addition, the RPA70 N terminus (1–211 aa) was confirmed to be required for G9a binding (Fig. S5E).

Fig. S5.

G9a, but not H3K9me2, interacts with RPA70. This is related to Fig. 5. (A) HCT116 cells were separately transfected with pcDNA3.1(+) or FLAG-G9a. The Flag immunoprecipitates were separated by SDS/PAGE and stained with CBB. The entire pcDNA3.1(+) or G9a lanes were excised from the gel and analyzed by mass spectrometry separately. The table provides summaries of the proteins identified by mass spectrometry. Letters in boldface type indicate the bait proteins. (B) HCT116 cells were transiently transfected with pcDNA3.1(+) or FLAG-G9a. The cell extracts were precipitated using M2 beads, and the immunoprecipitates were analyzed using anti-RPA70 and anti-RPA32 antibodies. (C) Nuclear proteins were extracted from HeLa cells and subjected to immunoprecipitation using anti-RPA32 or anti-G9a antibodies. The arrows indicate the corresponding bands. (D) Recombinant G9a (the GST-tag of GST-G9a was cut with human rhinovirus 3C protease) and GST-RPA32 were incubated in vitro in GST pull-down buffer. Western blotting was performed to detect G9a levels. CBB staining was performed to detect GST-tagged proteins. The arrow indicates the corresponding bands. (E) Bacterially purified GST-tagged RPA70 fragments were incubated with recombinant G9a. (Upper) The schematic structure of RPA70. FL: full length. F, A, B, and C are four DNA-binding domains of RPA70. CBB staining was performed to detect GST-tagged proteins. (Lower) The arrows show the corresponding bands. (F) Nuclear proteins from HCT116 cells treated with or without VP16 were incubated with biotin-labeled H3K9me2 peptide. Western blotting was performed to detect possible H3K9me2-interacting proteins. The interaction between HP1γ and H3K9me2 or between G9a and H3K9me2 was used as a positive control.

Fig. 5.

G9a interacts with RPA in vivo and in vitro. (A) HCT116 cells were transfected with FLAG-G9a for 48 h followed by VP16 treatment. The FLAG immunoprecipitates were analyzed using anti-RPA70 antibodies. (B) Nuclear proteins were extracted from HCT116 cells, and Co-IP was performed to detect the interaction between G9a and the RPA complex. (C) Soluble nucleoplasm proteins and soluble chromatin proteins were exacted from HCT116 cells with or without VP16 treatment. Co-IP was performed as described in B. (D) HCT116 cells were cotransfected with GFP-G9a and FLAG-RPA70 or GFP-G9a and FLAG-RPA32. The GFP immunoprecipitates were analyzed using anti-FLAG or anti-GFP antibodies. (E) HCT116 cells were cotransfected with GFP-G9a and FLAG-RPA70 (a), GFP-G9a and FLAG-RPA32 (b), or GFP-G9a and Flag-GLP, respectively. The FLAG immunoprecipitates were analyzed using anti-FLAG or anti-GFP antibodies. The interaction between FLAG-GLP and GFP-G9a was used as a positive control, and the GLP bands of a and b are the same. (F) HCT116 cells were separately transfected with pcDNA3.1(+) FLAG-G9a (WT), FLAG-G9a (S211A), or FLAG-G9a (S211D). The FLAG-tagged immunoprecipitates were subjected to Western blotting using an anti-RPA70 antibody. “FLAG” indicates FLAG-tagged G9a WT or its mutants. (G) Bacterially purified GST or GST-RPA70 was coincubated with recombinant G9a (the GST-tag of GST-G9a was cut with human rhinovirus 3C protease). Western blotting was performed to detect G9a protein levels and Coomassie Brilliant Blue (CBB) staining was performed to detect GST or GST-RPA70 levels. The arrow indicates the corresponsive protein bands. (H) Bacterially purified GST-G9a fragments were coincubated with recombinant RPA70 (the GST-tag of GST-RPA70 was cut with HRV3C protease). (Upper) The schematic structure of G9a. ANK repeats: ankyrin repeats domain; E: Glu-rich region; Post: Post-SET domain; Pre: Pre-SET domain; SET: catalytic domain. Western blotting was performed to detect RPA70 protein levels, and CBB staining was performed to detect GST or GST-tagged proteins. The arrows indicate the corresponding protein bands. (I) HCT116 cells were cotransfected with GFP-G9a (WT) and FLAG-RPA70, GFP-G9a (1–350 aa) and FLAG-RPA70, or GFP-G9a (1–630 aa) and FLAG-RPA70. Co-IP was performed to detect the interactions between the GFP-tagged G9a fragments and FLAG-RPA70.

H3K9me2 interacts with the BARD/BRCA1 complex through HP1γ (40), and therefore, it should be ruled out that the function of G9a in response to DNA damage may be mediated by H3K9me2. To assess this possibility, a peptide pull-down assay was performed to detect H3K9me2-interacting proteins. The assay indicated that RPA70, Rad51, BRCA1, and CtIP did not interact with the biotin-labeled H3K9me2 peptide, whereas G9a and HP1γ (41, 42), two positive controls, were pulled down (Fig. S5F). These results suggest that G9a, but not H3K9me2, binds to RPA70 in response to DNA damage.

G9a Depletion Impairs Loading of the RPA and Rad51 to DSBs.

Subsequently, the RPA and Rad51 foci formation in HCT116 G9a-WT or G9a-KO cells was tested to assess whether G9a can modulate RPA function upon DNA damage. As shown in Fig. 6 A and B, IR exposure or VP16 treatment resulted in a significantly higher number of RPA and Rad51 foci in G9a-WT cells compared with that in G9a-KO cells (P < 0.05). In addition, the cell-cycle distribution was unaltered, indicating that no specific cell-cycle stage is perturbed following G9a depletion (Fig. S6A). The same experiments were repeated and confirmed in HeLa cells (Fig. S6B). The recruitment of DNA-responsive proteins upstream of RPA, such as BRCA1, 53BP1, and MRE11, was not impaired by G9a deficiency (Fig. S6C). Moreover, the recruitment of Cherry-tagged RPA70 to the IR path was significantly delayed in HCT116 G9a-KO cells compared with HCT116 G9a-WT cells (Fig. S6D). Furthermore, Western blot analysis of RPA70 and Rad51 protein levels on chromatin confirmed a moderate decrease in their abundance as a result of G9a deficiency under normal conditions. However, a marked decrease in RPA70 and Rad51 accumulation on chromatin was observed following IR, VP16, or CPT treatment (Fig. 6 C and D and Fig. S6E). RPA70 recruitment to chromatin upon hydroxyurea (HU) treatment was also impaired by G9a knockout, suggesting that G9a may also have a role in the cellular response to DNA replication stress (Fig. S6F). Total protein levels of RPA70, RPA32, Rad51, and other DDR proteins were also not affected by G9a deficiency following VP16 treatment (Fig. S6G). These data suggest that deficiency of G9a in cancer cells impairs RPA and Rad51 foci formation.

Fig. 6.

G9a depletion impairs loading of the RPA and Rad51 to DSBs. (A) G9a-WT or G9a-KO HCT116 cells were exposed to IR or VP16 and then recultured in normal conditions for 1 h before labeling with an anti-RPA or anti-Rad51antibody. (B) Quantification of RPA-positive or Rad51-positive cells. The RPA-positive or Rad51-positive cells refer to cells with ≥5 RPA or Rad51 foci, respectively. The data represent the means ± SD (n ≥100) from three independent experiments. (C) G9a-WT or G9a-KO HCT116 cells were exposed to IR and recultured under normal conditions for 3 min and 1, 3, and 6 h. Chromatin-bound proteins were subjected to Western blotting with the indicated antibodies. (D) G9a-WT or G9a-KO HCT116 cells were treated with or without VP16 or CPT, and the chromatin-bound proteins were subjected to Western blotting with the indicated antibodies. (E) DR-U2OS or EJ5-U2OS cells were transfected with an siRNA against G9a. Twenty-four hours after transfection, the cells were transfected with I-SceI plasmid. At 48 h after transfection, the cells were harvested and assayed for GFP-positive signals by fluorescence-activated cell sorting (FACS) analysis. The data represent the means ± SD from three independent experiments. Western blotting was performed to detect G9a knockdown efficiency.

Fig. S6.

G9a depletion impairs loading of the RPA and Rad51 to DSBs. This figure is related to Fig. 6. (A) G9a-WT or G9a-KO HCT116 cells were analyzed by flow cytometry. (B) HeLa cells were transfected with an siRNA against G9a for 48 h and treated with or without IR or VP16 before being fixed and labeled with the indicated antibodies. The arrows show the cells with G9a knockdown. (C) G9a-WT or G9a-KO HCT116 cells were mixed and cultured in the same plates. After IR exposure, the cells were fixed and stained with indicated antibodies. The arrows indicate the G9a knockout cells. (D) G9a-WT or G9a-KO HCT116 cells transfected with m-Cherry-RPA70 were irradiated with a 365-nm UV laser beam. Images were collected every 20 s after laser microirradiation. The arrows indicate the irradiation path. (E) G9a-WT or G9a-KO #2 HCT116 cells were treated with or without VP16 or CPT, and the chromatin-bound proteins were subjected to Western blotting with the indicated antibodies. (F) G9a-WT or G9a-KO HCT116 cells were treated with or without HU (4 mM), and the chromatin-bound proteins were subjected to Western blotting with the indicated antibodies. (G) Whole-cell proteins from G9a-WT or G9a-KO HCT116 cells were subjected to Western blotting with the indicated antibodies. (H) Nuclear proteins were separately extracted from G9a-WT or G9a-KO HCT116 cells and subjected to a coimmunoprecipitation assay to detect the interaction between RPA70 and RPA32. (I) The RPA complex was purified by M2 beads from HCT116 cells with transfection of FLAG-RPA70. Then biotin-labeled ssDNA was incubated with RPA complex in the absence or presence of recombinant G9a. Western blotting was performed to detect RPA70, RPA32, and G9a levels. The biotin-labeled ssDNA sequence is 5′ biotin-TTGTAAAACGCGGCCAGTGAATTCATCATCAATATTCCTTTTTTGGC AGGCGGTGTTAATACTGCCGCCTAATCG 3′. The Upper band of RPA70 indicates the exogenous RPA70 and the Lower band indicates the endogenous RPA70. (J) Biotin-labeled ssDNA was incubated with RPA complex in the absence or presence of recombinant G9a (S211A) or G9a (S211D). Western blotting was performed as described in I.

To explore the mechanisms by which loading of RPA and Rad51 to DSBs is impaired as a result of G9a deficiency, the ssDNA-binding activity of RPA was investigated in vitro. In the absence of G9a, low levels of RPA could be precipitated with biotin-labeled ssDNA, but this interaction was markedly increased in the presence of functional G9a (Fig. S6I). G9a deficiency did not affect the binding activity of RPA70 to RPA32 (Fig. S6H). These results indicate that loss of G9a disturbs the loading of RPA and Rad51 to chromatin in response to DSBs, probably by enhancing the stability of the RPA–ssDNA complex.

We next investigated the downstream events of RPA–ssDNA complex formation. The phosphorylation levels of Chk1 and RPA32, which is activated by RPA-coated ssDNA (43, 44), were decreased as a result of G9a deficiency (Fig. S7A). And the percentage of the cells in the G2/M phase decreased ∼10% at 18 h post IR or VP16 withdrawal in G9a-KO cells compared with that in G9a-WT cells (Fig. S7B). These results indicate that G9a deficiency disturbs the activation of the G2/M DNA damage checkpoint.

Fig. S7.

G9a depletion impairs G2/M checkpoint activation and HR. This figure is related to Fig. 6. (A) G9a-WT or G9a-KO HCT116 cells were treated with or without VP16 or CPT treatment. Soluble nuclear extracts were subjected to Western blotting using the indicated antibodies. (B) G9a-WT or G9a-KO HCT116 cells were exposed to IR or VP16, allowed to culture under normal conditions for the indicated time course, and analyzed by flow cytometry. The percentage of G2/M cells is shown. (C) DR-U2OS or EJ-U2OS cells were transfected with an siRNA against G9a for 48 h followed by flow cytometry.

Next, the DNA repair efficiency was analyzed in DR-U2OS or EJ-U2OS cells—two well-known cell lines for detecting HR or NHEJ efficiency, respectively (37, 45). As shown in Fig. 6E, G9a knockdown by siRNA significantly decreased the efficiency of HR (according to GFP expression following I-SceI expression) (P < 0.05), whereas it had little effect on NHEJ. In addition, G9a knockdown did not alter the cell-cycle profiles of DR-U2OS or EJ-U2OS cells (Fig. S7C). Collectively, G9a has an important role in G2/M checkpoint activation and HR in response to DSBs.

G9a Is Recruited to Chromatin by RPA.

RPA-coated ssDNA is a platform for the recruitment of many DNA damage-responsive proteins to chromatin (9); thus we sought to explore whether RPA also recruits G9a to chromatin. As shown in Fig. S8A, RPA70 knockdown impeded the accumulation of G9a on chromatin in response to VP16 treatment. In addition, we monitored the recruitment of G9a to DNA damage stripes in cells depleted in CtIP, an upstream factor of DNA end resection and RPA loading (12). As shown in Fig. S8 B and C, CtIP depletion markedly impaired GFP-G9a and Cherry-RPA70 recruitment to DNA damage stripes, suggesting that G9a functions downstream of CtIP. Taken together, these results support the idea that G9a is recruited to chromatin by the RPA complex.

Fig. S8.

G9a is recruited to chromatin by RPA complex. (A) HCT116 cells were transfected with an siRNA against RPA70 for 48 h followed by treatment with VP16. Dt or Chr proteins were extracted and subjected to Western blotting using the indicated antibodies. (B) Whole-cell lysates from CtIP^+/+ or CtIP^−/− HCT116 cells were analyzed by Western blotting to detect CtIP protein levels. (C) CtIP^+/+ or CtIP^−/− cells were transfected with m-Cherry-RPA70 or GFP-G9a and irradiated with a 365-nm UV laser beam. Images were collected every 20 s after irradiation. The arrows indicate the irradiation path.

Interaction Between G9a and RPA Is Required for DNA Damage Repair.

To confirm the results described above, a rescue experiment was performed whereby G9a (WT) or different G9a mutants were reintroduced into G9a-KO HCT116 cells, and RPA recruitment and cell survival were analyzed. Reintroduction of G9a (WT), G9a (S211D), or G9a (1–630 aa) was sufficient to recruit RPA70 to damaged DNA stripes in G9a-KO cells, but G9a (S211A) failed to rescue RPA70 recruitment (Fig. 7A), suggesting that both the phosphorylation of G9a at Ser211 and the N terminus of G9a are indispensable for RPA70 recruitment to damaged DNA stripes. We also performed Western blotting in HCT116 G9a-KO cells to confirm this result. As shown in Fig. 7 B and C, overexpression of FLAG-G9a (WT), FLAG-G9a (S211D), or FLAG-G9a (1–630 aa), but not FLAG-G9a (S211A), in G9a-KO cells increased the recruitment of RPA70 and Rad51 to chromatin in response to VP16 treatment. In addition, the decreased efficiency of HR in G9a knockdown cells was rescued in cells in which siRNA-resistant G9a (WT) or G9a (S221D) were overexpressed, but not in cells where G9a (S211A) was overexpressed (Fig. 7D). Consistently, colony formation assay confirmed that the cell-survival rate increased when G9a (WT) or G9a (S211D) were re-expressed in G9a-KO cells, but not G9a (S211A) (Fig. 7E). Re-expression of exogenous G9a without the SET catalytic domain [G9a (D960-1210)] failed to increase cell survival in G9a-KO cells (Fig. S9A), suggesting that G9a methylation activity is required for cell survival under basal conditions. Overall, these data support that the Ser211 phosphorylation of G9a or the N terminus of G9a, which facilitates the interaction between G9a and RPA, is required for DNA damage repair.

Fig. 7.

The interaction between G9a and RPA70 is required for DNA damage repair. (A) HCT116 G9a-KO cells were transfected with m-Cherry-RPA70 alone or m-Cherry-RPA70 together with GFP-G9a (WT), GFP-G9a (S211A), GFP-G9a (S211D), or GFP-G9a (1–630 aa) before irradiation with a 365-nm UV laser beam. Images were collected every 20 s after laser microirradiation. The arrows indicate the irradiation path. (B) HCT116 G9a-KO cells were transfected with pcDNA3.1(+) FLAG-G9a (WT), FLAG-G9a (1154YF), FLAG-G9a (D960-1210), or FLAG-G9a (1–630 aa) for 48 h and then treated with or without VP16. Chromatin proteins were extracted for Western blotting analysis. (C) HCT116 G9a-KO cells were transfected with pcDNA3.1(+) FLAG-G9a (WT), FLAG-G9a (S211A), or FLAG-G9a (S211D) for 48 h and then treated with or without VP16. Chromatin proteins were extracted for Western blot analysis. (D) siRNA-resistant G9a (WT), G9a (S211A), or G9a (S211D) were reintroduced into G9a knockdown DR-GFP U2OS cells. At 48 h after transfection, the cells were harvested and assayed for GFP-positive signals by FACS analysis. The data represent the means ± SD from three independent experiments. (E) HCT116 G9a-KO cells were transfected with G9a (WT), G9a (S211A), or G9a (S211D) and subjected to a colony formation assay. The data represent the means ± SD from three independent experiments. Student’s t test (two-tailed): WT versus pcDNA3.1(+) P < 0.05; WT versus S211A, P < 0.05; S211D versus pcDNA3.1(+) P < 0.05; S211D versus S211A, P < 0.05.

Fig. S9.

The interaction between G9a and RPA is required for DNA damage repair. This figure is related to Fig. 7. (A) Flag-G9a (WT), Flag-G9a (D960-1210), or pcDNA3.1(+) were reintroduced into HCT116 G9a-KO cells, and then colony formation assays were performed. The clone numbers under each condition are shown. The clone numbers of HCT116 cells with re-expressed Flag-G9a (WT) or Flag-G9a (D960-1210) were the same as that in Fig. 7C. (B) A working model for this study is presented and described in Discussion.

Discussion

The present study has identified G9a as a direct and positive regulator of HR. Upon DNA damage, G9a is recruited to chromatin as a result of CK2-mediated phosphorylation. Chromatin-enriched G9a can then interact with RPA to permit RPA and Rad51 foci formation. Loss of G9a results in attenuated G2/M checkpoint activation and reduced efficiency of HR. Therefore, a deficiency in G9a will impair cancer-cell survival (Fig. S9B).

G9a is a well-known transcriptional repressor (46), but also has functions that are independent of its methyltransferase activity. For example, G9a promotes p53-dependent activation of Puma by interacting with histone acetylase p300/CBP (47), and, in response to glucocorticoid, G9a acts as a platform for the recruitment of gene coactivators (48). Here, we have identified a role for G9a whereby G9a directly interacts with RPA to help regulate HR in cancer cells.

It is well recognized that modifiers of histone methylation have pivotal roles in the DDR. For example, SETD2 mediates H3K36me3 on DSBs, which recruits LEDGF to chromatin to facilitate DNA end resection (49). Down-regulation of SUV39H1 methyltransferase activity leads to heterochromatin relaxation and genomic instability in response to DNA damage in cancer cells (50). In our study, we found that the histone methyltransferase G9a is involved in DNA damage repair. However, we noted a conflicting phenomenon that the reduction of γ-H2AX foci post IR in the cancer cells tested (Fig. S1 F and H) is much faster than what has been previously reported (51, 52). Although the exact mechanism by which the differing γ-H2AX kinetics occurred in different studies is still unknown, we aimed to examine the kinetics of γ-H2AX post IR in normal human cells. Interestingly, we found that it takes >12 h for the recovery of γ-H2AX signals post IR to basal levels in mouse embryonic fibroblasts (Fig. S1I), suggesting that the γ-H2AX kinetics may vary depending on the cell line used.

Although we could not detect G9a recruitment to IR-induced foci, we found that G9a is readily recruited to DNA damage sites after laser microirradiation. Recent studies have recognized that multiphoton treatment, such as laser microirradiation, causes damage to DNA and proteins (53–55). To address whether G9a is involved in DNA damage or protein damage, we observed the recruitment of GFP-G9a to the IR path following laser microirradiation in cells with or without 5-bromo-2′-deoxyuridine (BrdU) treatment. As shown in Fig. S2C, GFP-G9a recruitment to DNA damage sites was significantly enhanced in cells pretreated with BrdU compared with non-BrdU preincubated cells. Considering that BrdU treatment sensitizes cancer cells only to DNA damage but not to protein damage, this result suggests that G9a is implicated in the response to DNA damage and thus recruited to the IR path.

The recruitment of G9a to chromatin is not dependent on the canonical PIKK family members ATM, ATR, and DNA-PK (56) but instead on the kinase CK2. Other substrates of CK2 have already been identified as being important for DNA damage repair. For example, within a few minutes of DNA damage, phosphorylation of HP1-β at Thr51 by CK2 compromises its chromatin-binding activity, leading to HP1-β release from chromatin to facilitate H2AX phosphorylation and initiation of the DDR (57). CK2-mediated phosphorylation of Rad51 at Thr14 triggers its direct binding to NBS1, which promotes Rad51 recruitment to sites of DNA damage and facilitates HR (58). Thus, compared with the canonical PIKK family substrates, the role of CK2 in mediating substrate phosphorylation is also critical for promoting DNA damage repair.

The repetitive ankyrin domain of G9a is essential for H3K9me1 and H3K9me2 binding (41); however, under most conditions, the recruitment of G9a to chromatin is not dependent on its methylation-binding domain. In nonneuronal cells, the transcriptional repressor NRSF recruits G9a to NRSF target genes (59). In response to viral infection, the DNA-binding protein PRDI-BF directs G9a to the IFN-β promoter to silence the surrounding chromatin in U2OS osteosarcoma cells (60). In our study, phosphorylation at Ser211 residue was required for G9a recruitment by RPA to chromatin. In addition, the interaction between G9a and RPA70 was notably impaired in vivo upon mutation of G9a S211 to A211 (Fig. 5F). However, as demonstrated by in vitro pull-down assay, both recombinant G9a (1–350 aa) and G9a (350–630 aa) portions could bind to RPA70 (Fig. 5H), suggesting that Ser211 phosphorylation is not required for G9a binding to RPA70 in vitro. This conflicting phenomenon has also been reported in the binding of G9a to CDYL (61), which suggests that the interaction between G9a and RPA70 may be regulated in vivo by combinatorial posttranslational modifications to both proteins. Therefore, it is possible that G9a Ser211 phosphorylation may lead to the conformational changes of G9a and thus facilitate its binding to RPA70.

RPA is a key eukaryotic ssDNA-binding protein that participates in a variety of DNA biological processes, including DNA replication and DNA repair (9, 62). Other DNA-binding proteins compete with RPA to bind ssDNA, which is considered to be the main way by which RPA–ssDNA formation is regulated. For example, BRCA2-mediated Rad51–ssDNA nucleoprotein formation dissociates RPA from ssDNA at DSBs to promote DNA strand exchange (63). Heterogeneous nuclear ribonucleoprotein A1 can also displace RPA from telomeric ssDNA after DNA replication to maintain the length of telomeres (64). In addition, several proteins regulate the dissociation of RPA from ssDNA by directly binding to RPA70, including SLFN11, which can destabilize the RPA–ssDNA complex (65). This study found that G9a was directly associated with the N terminus (1–210 aa) of RPA70 and enhanced RPA loading to chromatin, likely by increasing its ssDNA-binding capacity (Fig. S6I). Mutation of G9a S211 to D211, however, did not markedly change the binding capacity of RPA to ssDNA compared with G9a (S211A) in vitro (Fig. S6J), suggesting that both G9a WT and G9a mutants could equally enhance RPA affinity to ssDNA in vitro. Considering that the N terminus of RPA70 interacts with numerous proteins (66), it is also possible that G9a promotes RPA loading to ssDNA by increasing or decreasing the binding ability of other factors to RPA70. The exact mechanism by which G9a promotes RPA foci formation remains to be investigated in further studies.

In the present study, G9a, but not its histone substrate H3K9me2, was found to interact with RPA to facilitate DNA repair via HR. Interestingly, previous studies have suggested that H3K9me2 is required for HR. The levels of H3K9me2 increase at DSBs and H3K9me2-containing H3 peptide has been shown to bind to HA-tagged BRCA1 in vitro (67). In addition, the interaction between H3K9me2 and HP1 is essential for BRCA1–BARD complex retention at sites of damaged DNA, which can be abolished upon application of the specific G9a inhibitor UNC0638 (40). Consistently, we also observed increased levels of H3K9me2 at chromatin in response to DNA damage, and this was eliminated by G9a depletion (Fig. S4 D–F). However, the interaction between H3K9me2 and BRCA1 was not detected in our in vitro peptide pull-down assays (Fig. S5F). This difference between our study and the previous report (67) may be because the interaction between H3K9me2 and endogenous BRCA1 is weak, causing it to dissociate during the peptide pull-down. As the levels of H3K9me2 were significantly reduced upon G9a knockout in basal conditions, overexpression of the G9a N terminus failed to rescue the cell survival in G9a-KO cells (Fig. S9A). Taken together, the histone methylation activity of G9a is confirmed to be dispensable for G9a’s function in the response to DSBs, but it is indispensable for cancer-cell survival.

In conclusion, this study has identified a regulatory pathway based on a CK2–G9a–RPA axis that is required for RPA foci formation and HR upon DNA damage. This function of G9a in DNA repair provides an aspect by which to understand its role in cancer development and progression and may be exploited in the design of new cancer therapeutics.

Materials and Methods

Laser Microirradiation.

Cells grown in culture dishes with a thin glass bottom were transiently transfected with GFP-tagged or m-Cherry-tagged plasmids for 24–36 h and then irradiated with a 365-nm pulsed nitrogen UV laser (16 Hz pulse, 50% laser output) generated using a MicroPoint photo-stimulation system (Andor). This system was directly coupled to the epifluorescence path of a Nikon A1 confocal imaging system, and time-lapse images were captured every 20 s for 10 min.

Chromatin Protein Extraction.

Cells were lysed in buffer I (10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.15% Nonidet P-40, 1% protease inhibitor mixture, and 1 mM DTT) for 10 min. After centrifugation at 12,000 × g for 30 s, the cell pellets were washed twice in PBS and then lysed in buffer II (3 mM EDTA, 0.2 mM EGTA, 1% mixture, and 1 mM DTT) for 30 min. After centrifugation at 12,000 × g for 3 min, the supernatant was assumed to contain soluble nucleoproteins (Dt), and the pellets were assumed to contain the chromatin fraction (Chr).

SI Materials and Methods

Cell Culture.

Cells were grown in DMEM or McCoy’s 5A with 10% (vol/vol) FBS and the appropriate amount of penicillin/streptomycin in a 37 °C incubator with a humidified, 5% CO₂ atmosphere.

Plasmids Construction.

The G9a full-length gene (isoform a) or fragments were separately subcloned into pEGFP-C1, p3xFLAG-CMV-10, or pGEX-6p1 vectors. CK2α was amplified from a cDNA library of HCT116 cells and cloned into p3xFLAG-CMV-10. RPA32 or RPA70 were separately amplified and cloned into p3xFLAG-CMV-10, pGEX-6p1, or m-Cherry-N1 vectors. G9a or CK2α mutants were generated using a site-directed mutagenesis kit (Stratagene). Transient and stable transfections of these plasmids were performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol.

Antibodies.

The antibodies used were anti–GFP-tag; anti–Flag-tag (Sigma-Aldrich); anti–GST-tag (Applygen); anti-G9a (for confocal, Cell Signaling Technology; for Western blotting and coimmunoprecipitation, Sigma-Aldrich); anti–pan-serine, anti–pan-methyl, anti-H3K9me2, anti-H3, and anti-RPA32 (for confocal, Abcam); anti-BRCA1, anti-Rad51, anti-actin, anti–β-tubulin, anti-p53S15ph, anti-RPA32 (for Western blotting, Santa Cruz); anti–phospho-Histone H2AX (Ser139) (for confocal microscopy, Millipore; for Western blot, Cell Signaling Technology); anti–p-RPA32, anti-CK2α′, anti-GLP (Bethyl); anti-CK2α, anti-RPA70, anti–pan-threonine, anti–pan-tyrosine, anti–p-Chk1(s354), anti-Chk1 (Cell Signaling Technology), and anti-53BP1 (Novus Biologicals).

Generation of G9a Knockout Cell Lines.

HCT116 cells were cotransfected with CRISPR-Cas9 plasmids and two small-guided RNAs (sgRNAs) using polyethylenimine (purchased from Polysciences). The two sgRNA sequences designed to target the human G9a (EHMT2) genes are sgRNA a: ATGAGTGGTGTAGCCCCTAC and sgRNA b: GCAGGGTTTCTTCACTA CGA. Twenty-four hours after transfection, cells were cultured in selective medium containing puromycin (1–2 μg/mL) for 2 wk. Surviving colonies were then selected and expanded. The G9a knockout cells were then analyzed by Western blotting with a G9a-specific antibody and by DNA sequencing using the primers Forward a: GACACCCCTCGTAGTGAAGAAACCC; Reverse a: GAAAGAAGGCAAGAGTCAG; Forward b: TCTCCTGAGCCCTTAGCTCTG; and Reverse b: GAGGCTGGAGATGAGGGGCC.

Immunofluorescence.

Cells were cultured on glass slides to ∼80% confluency. After appropriate treatment, cells were fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100. The slides were then blocked in 3% BSA in PBS for 30 min and exposed to primary antibody (1:200 dilution for all antibodies) overnight at 4 °C. The cells were then washed three times in PBS and then exposed to a secondary antibody conjugated to FITC/TRITC for 1 h at room temperature in the dark. The slides were then embedded in DAPI for 2 min at room temperature. Immunofluorescence images were obtained using an Olympus FV1000 inverted confocal IX81 microscope.

Colony Formation.

Cells exposed to IR or treated with VP16 were washed in PBS three times, digested, and then counted before seeding in six-well plates in an equal number. After culture in a drug-free medium for 2 wk, the cells were fixed using methanol and stained with crystal violet to identify visible colonies. The number of colonies was calculated using Image J.

RNAi Assay.

Sequences of RNAi oligonucleotides are as follows: siCTR—UUCUCCGAACGUGU; CACGUTT, siG9a #1— GCUCCAGGAAUUUAACAAGAU; siG9a #2—CCAUGCUG UCAACUACCAUGG; siCK2α: GAUGACUACCAGCUGGUUC; siCK2α′: CAGUCUGAGGAGCCGCGAG; and siRPA70: AACACUCUAUCCUCUUUCAUG. All RNAi oligonucleotides were purchased from Shanghai GenePharma. These RNAi oligonucleotides were transfected into cells using the Lipofectamine 2000 transfection kit (Invitrogen) according to the manufacturer’s instructions.

ChIP Assay.

Cells were fixed with formaldehyde and lysed with lysis buffer (50 mM Tris·HCl, pH 8.0, 5 mM EDTA, 1% SDS). After sonication, the supernatant was collected by centrifugation and precleared in dilution buffer (20 mM Tris·HCl, pH 8.0, 2 mM EDTA, 150 mM NaCl, 1% Triton X-100) with protein G or A Sepharose and salmon sperm DNA. The precleared samples were incubated with the indicated antibody and protein G or A Sepharose. The beads were washed and heated to 65 °C to reverse the formaldehyde cross-links. DNA was purified and real-time PCR was performed using the following primers: 5′-TCTTCTTCAAGGACGACGGCAACT-3′ (sense) and 5′-TTGTAGTTGTACTCCAGCTTGTGC-3′ (antisense).

Coimmunoprecipitation.

Cells were lysed in lysis buffer [1% Nonidet P-40, 150 mM NaCl, 20 mM Tris⋅HCl, pH 8.0, 10% glycerol, 2 mM EDTA, 1% EDTA-free protease, and phosphatase inhibitor mixtures (Roche)] on ice for 30 min. After centrifugation at 12,000 × g for 15 min at 4 °C, 2 μg of the indicated antibody was added to the supernatant and incubated at 4 °C overnight. Then, 30 μL of protein G or A Sepharose slurry (GE Healthcare) was added, and the sample was incubated for a further 2 h at 4 °C. The beads were washed in Nonidet P-40 buffer three times. The precipitated components were analyzed by Western blotting.

Nuclear Protein Coimmunoprecipitation.

Cells were lysed with buffer A (10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.15% Nonidet P-40, 1% protein inhibitor mixture, and 1 mM DTT) for 10 min. After centrifugation, the deposits were washed twice in PBS and then lysed in buffer B (20 mM Hepes, pH 7.9, 40 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 0.5% Nonidet P-40, 1% Mixture) for 20 min, shaking for 10 s at 5-min intervals. After centrifugation at 12,000 × g for 10 min at 4 °C, the supernatant was diluted with buffer B′ (20 mM Hepes, pH 7.9, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1% Mixture) at a volume ratio of 1: 1 (vol/vol) and precleared with protein A or G Sepharose for 2–4 h. After centrifugation at 2,000 × g for 5 min at 4 °C, 2 μg of the indicated antibody was added to the supernatant and incubated at 4 °C overnight. After centrifugation at 2,000 × g for 5 min at 4 °C, 30 μL of protein G or A Sepharose was added and incubated for a further 2 h at 4 °C. The beads were washed with buffer B three times. The precipitated components were analyzed by Western blotting.

In Vitro Phosphorylation Assay.

Reactions were performed in a total volume of 40 μL in 10 mM Tris⋅HCl, pH 7.4, 150 mM NaCl, 10 mM MgCl2, 0.5 mM DTT, 10 μm cold ATP, 1 μg recombinant G9a (WT) or G9a (S211A), and 100–200 ng CK2 (WT) or CK2 (K69M). The reaction mixtures were incubated at 37 °C for 1 h.

Protein Purification.

GST-fusion proteins were expressed in the bacterial cell strain BL21 with isopropyl β-d-1-thiogalactopyranoside, purified using glutathione-Sepharose 4B beads (GE Healthcare), and then washed in TEN buffer (20 mM Tris⋅HCl, pH 7.4, 0.1 mM EDTA, and 100 mM NaCl).

GST Pull-Down.

The GST-fusion proteins (“bait” and “prey” proteins) were purified as described in Protein Purification. For GST pull-down, the GST tag of the prey proteins was cut using HRV 3C protease (Takara) according to the manufacturer’s protocol. The bait and prey proteins were then incubated at 4 °C overnight in GST-binding buffer (50 mM Tris⋅HCl, pH 7.4, 150 mM NaCl, 0.05% Nonidet P-40). The beads were washed three times with GST-binding buffer and boiled in 2× SDS loading buffer, and the proteins were analyzed by Western blotting with the indicated antibodies.

Biotin–ssDNA Pull-Down.

A total of 8 μL biotin–ssDNA and 30 μL streptavidin Sepharose beads were preincubated in PBS with rotation at 4 °C overnight, and then the beads were absorbed by magnetism. At the same time, FLAG-tagged RPA complex and recombinant G9a (without GST-tag) were incubated in GST-binding buffer at 4 °C overnight. The biotin-labeled ssDNA was incubated with GST-binding buffer containing FLAG-tagged RPA complex and recombinant G9a for 4–6 h at 4 °C. The beads were washed in GST washing buffer (50 mM Tris⋅HCl, pH 7.4, 150 mM NaCl, 0.5% Nonidet P-40) three to four times, and the immunoprecipitates were analyzed by Western blotting.

BrdU Incorporation.

Cells were cultured to ∼80% confluency and treated with 10 μM BrdU for 30 min. The cells were collected and fixed with 75% ethanol at 4 °C under rotation overnight. After two washes with PBS, the DNA was denatured in 2N HCl with 0.5% Triton X-100 at room temperature for 30 min. The samples were then centrifuged at 500 × g for 3 min, and the cells were washed twice with PBS. The cells were then incubated in PBS containing 0.5% Tween 20, 1% BSA, and anti-BrdU antibodies at 4 °C with a rotation overnight. The cells were washed twice in PBS and exposed to a secondary antibody conjugated to FITC at room temperature for 30 min. The cells were washed before incubation with RNase A (250 μg/mL) at 37 °C for 30 min and then stained with propidium iodide (5 μg/mL). The cells were analyzed by flow cytometry.

Statistical Analysis.

Statistical significance was evaluated by paired Student’s t test: *P < 0.05, **P < 0.01, and ***P < 0.001.