Expression of BRC Repeats in Breast Cancer Cells Disrupts the BRCA2-Rad51 Complex and Leads to Radiation Hypersensitivity and Loss of G2/M Checkpoint Control*

BRCA2 is a breast tumor suppressor with a potential function in the cellular response to DNA damage. BRCA2 binds to Rad51 through its BRC repeats. In support of the biological significance of this interaction, we found that the complex of BRCA2 and Rad51 in breast cancer MCF-7 cells was diminished upon conditional expression of a wild-type, but not a mutated, BRC4 repeat using the tetracycline-inducible system. Cells expressing a wild-type BRC4 repeat showed hypersensitivity to γ-irradiation, an inability to form Rad51 radiation-induced foci, and a failure of radiation-induced G2/M, but not G1/S, checkpoint control. These results strongly suggest that the interaction between BRCA2 and Rad51 mediated by BRC repeats is critical for the cellular response to DNA damage.

BRCA2 is a breast tumor suppressor with a potential function in the cellular response to DNA damage. BRCA2 binds to Rad51 through its BRC repeats. In support of the biological significance of this interaction, we found that the complex of BRCA2 and Rad51 in breast cancer MCF-7 cells was diminished upon conditional expression of a wild-type, but not a mutated, BRC4 repeat using the tetracycline-inducible system. Cells expressing a wild-type BRC4 repeat showed hypersensitivity to ␥-irradiation, an inability to form Rad51 radiationinduced foci, and a failure of radiation-induced G 2 /M, but not G 1 /S, checkpoint control. These results strongly suggest that the interaction between BRCA2 and Rad51 mediated by BRC repeats is critical for the cellular response to DNA damage.
BRCA2 was cloned based on an analysis of mutations in families predisposed to breast cancer showing that a large percentage of the kindred had alterations within this locus (1,2). The expression pattern of BRCA2 is remarkably similar to that of BRCA1 (3)(4)(5), with highest levels in the testis, thymus, and ovaries (5). At the cellular level, expression is regulated in a cell-cycle dependent manner and peak expression of BRCA2 mRNA is found in S phase (6). These results suggest BRCA2 may participate in regulating cell proliferation.
Recent studies indicate that BRCA2 is important for the cellular response to DNA damage. Brca2-null mouse embryos are nonviable at a very early stage of development and blastocysts derived from these embryos are very sensitive to ␥-irradiation (7). Mouse embryonic fibroblasts predicted to express BRCA2 that is C-terminally truncated at amino acid 1492 also demonstrated sensitivity to DNA damaging agents, particularly methyl methanesulfonate and UV light (8). Furthermore, Capan-1, a human pancreatic cancer line, that expresses a 220-kDa C-terminally truncated BRCA2 protein, is hypersensitive to a panel of DNA damaging agents (9). Importantly, ectopic expression of wild-type, but not mutated, BRCA2 in Capan-1 cells restores resistance to treatment with MMS (10). These results provided convincing evidence that BRCA2 plays a critical role in the DNA repair process.
Interestingly, BRCA2 was shown to interact with Rad51 (7, 10 -12), a key protein in DNA recombinational repair. Human Rad51 encodes a 40-kDa protein with a structure related to the Escherichia coli recombination protein RecA (13) and mediates homologous DNA pairing and strand exchange (14,15). Similar to mBrca2, inactivation of mouse Rad51 results in an embryonic lethal phenotype, indicating that Rad51 protein is essential for development (16,17). Beyond serving as a DNA repair protein through its interactions with other Rad proteins including Rad52 and Rad54 (18), how Rad51 may participate in cell growth and development remains unclear.
While an association between BRCA2 and Rad51 is well documented, there is, nonetheless, some discrepancy concerning the regions of BRCA2 that bind to Rad51. It was reported that the C-terminal region of mouse Brca2 binds to mouse Rad51 (7). By contrast, we and others have previously shown that the BRC repeats located in exon 11 (amino acid 1009 -2083) of human BRCA2 bind to Rad51 (10,12). There are eight repeats in BRCA2 designated as BRC1 to BRC8 (Fig. 1A) (19,20). BRC1, BRC2, BRC3, BRC4, BRC7, and BRC8 are highly conserved and bind to Rad51, whereas BRC5 and BRC6 are less well conserved and do not bind to Rad51 (10,12). Whether the interaction between BRCA2 and Rad51 has biological significance remains completely unknown. In an effort to investigate this issue, we have used the tetracycline binary gene control system for conditional expression of a BRC4 repeat in breast cancer MCF-7 cells. In this communication, we have found that upon expression of a wild-type, but not a mutated, BRC4 repeat, the interaction between BRCA2 and Rad51 was reduced. Cells expressing a wild-type BRC4 repeat showed hypersensitivity to ␥-irradiation, an inability to form radiationinduced Rad51 nuclear foci, and a failure of radiation-induced G 2 /M checkpoint control. These results strongly suggest that the BRC repeats of BRCA2 are important for mediating the cellular response to DNA damage.

Isolation of Cell Clones with Inducible Expression of the BRC Re-
peat-To generate cell clones that express a GFP-BRC4 fusion protein, we employed the tetracycline-inducible expression system controlled by a tet-responsive promoter (24). A pUHD10-3-based plasmid was used to construct pUHD10-3/GFP-BRC plasmids that will express chimeric proteins containing GFP with a nuclear localization signal and a myc epitope fused to either a wild-type BRC4 or a mutated BRC4-M5. These two plasmids were separately co-transfected into MCF-7 cells with the second plasmid, pCHTV, which contains a hygromycin resistance gene and a cytomegalovirus-controlled tetracycline repressor-VP16 fusion transcription unit. Cell clones resistant to hygromycin were subsequently isolated and several of them were shown to express the wildtype or mutant GFP-BRC4 upon removal of tetracycline. The expression of GFP-BRC4 was further confirmed by immunoprecipitation with ␣-myc 9E10 monoclonal antibody (25) and immunoblotting analysis with a monoclonal ␣-GFP antibody (CLONTECH, Palo Alto, CA). Two stable lines of MCF-7 cells, WT-8 and MT-11, that conditionally express wild-type BRC4 and BRC4-M5 mutant, respectively, were established. Immunoprecipitations and Western Blotting-Immunoprecipitations were performed as described previously (10). Co-immunoprecipitations were performed similarly but with lysis buffer containing 180 mM NaCl. Antibodies specific for BRCA2 (10), human Rad51, Ab-1 (Oncogene Science, Cambridge, MA) and myc 1-9E10 monoclonal antibody (25) were used for the immunoblotting analysis according to standard procedure (10).
Clonogenic Survival Assay-Cells (WT-8 and MT-11) were seeded in identical plates at 5000 cells/plate in medium with tetracycline (1 g/ml). Expression of the wild-type or mutant GFP-BRC4 repeat was induced by removing tetracycline 24 h after seeding. Twenty four hours after induction of GFP-BRC4 expression, cells were then ␥-irradiated with 3 Gy. After incubation for 14 days, cells were fixed and stained with 2% methylene blue in 50% of ethanol for colony counting. Averages and standard deviations were determined from eight plates. Statistical analyses were performed with the programs InStat and InPlot (Graph-Pad Inc., San Diego, CA).
Cell Cycle Checkpoint Analysis-The G 1 /S checkpoint was determined according to the procedures described (27). Briefly, cells in logarithmic growth were mock-exposed or ␥-irradiated (12 Gy). After 24 h, cells were labeled with 10 M BrdUrd for 4 h and fixed for BrdUrd staining using a Cell Proliferation Kit (Amersham Pharmacia Biotech). BrdUrd-positive cells were quantified, and expressed as a fraction of the total cells. For the G 2 /M checkpoint, cells were irradiated to 3 Gy, fixed with 4% paraformaldehyde at indicated time and stained with DAPI for counting mitotic cells in prophase, metaphase, anaphase, and telophase (28). Alternately, cells were irradiated with 4 -16 Gy and processed for analysis of mitotic cells after 1 h.

RESULTS AND DISCUSSION
To systematically address the biological consequence of the interaction between BRCA2 and Rad51, amino acid residues of the first BRC repeat, BRC1, that are critical for Rad51-binding were first examined. BRC1 was subjected to biased PCR mutagenesis (21), and the mutated cDNAs were translationally fused to the GAL4 DNA-binding domain in the yeast vector, pAS1 (22), to generate a library of 2 ϫ 10 6 individual clones referred to as pAS/BRC1-ML. A reverse two-hybrid screen with negative selection was used to isolate clones that fail to bind Rad51 as described previously (23). Several mutations in BRC1 were identified that significantly reduced Rad51 binding in a yeast two-hybrid assay (Table I). BRC1-M1 is a mutation that changes a conserved threonine residue to alanine. BRC1-M2 and -M3 are changes in nonconsensus amino acids, and BRC1-M4 carries a double mutation at the two C-terminal BRC1 residues, the last residue of which is conserved. Interestingly, a familial mutation, G1529R, has been previously found in BRC4 (Breast Cancer Information Core). Specific Rad51 binding activity by BRC4 was also tested and found to be approximately three times stronger than BRC1 (Table I). Two BRC4 mutations, BRC4-M5, an analogous mutation to BRC1-M1, in which the conserved threonine at the third position is changed to an alanine, and BRC4-M6, which contains the G1529R mutation, were constructed and found to have reduced Rad51-binding (Table I). These results suggest that, despite their sequence conservation, the ability of BRC repeats to bind Rad51 varies, and is dependent on specific residues.
To determine the functional importance of the interactions between the BRC repeats of BRCA2 and Rad51, two stable lines of MCF-7 cells, WT-8 and MT-11, that conditionally express wild-type BRC4 and mutant BRC4-M5, respectively, were established (Fig. 1B). Tetracycline-responsive expression of the GFP-BRC4 fusion proteins in these two lines was clearly demonstrated by immunoprecipitation with ␣-myc antibodies and immunoblotting with ␣-GFP or ␣-Rad51 antibodies (Fig.  1C, top panel compare lanes 2 and 4 with 1 and 3). Rad51 is detected in the immunoprecipitates of wild-type, but not GFP-BRC4-M5 (Fig. 1C, compare lanes 4 with 2), indicating that the A randomly mutagenized pool of cDNAs encoding BRC1 repeats (amino acid 1003-1042) was cloned into the pAS1 vector and co-transformed along with pGAD-Rad51 into Mav203 cells. Four clones with DNA inserts that showed no detectable ␤-galactosidase activity in a yeast two-hybrid assay were isolated. Mutations with amino acid changes resulted from single nucleotide changes. Rad51 binding activity using the BRC4 in pAS1 vector was also tested in the assay. Note that this repeat has 3-fold higher activity compared with the BRC1 repeat. The T to A mutation identified above in the BRC1 repeat that abrogates Rad51 binding, and a familial mutation identified in BRC4 (G1529R) was introduced into the BRC4 and tested in the yeast-two hybrid assay. Both of the mutations significantly reduced, but the G to R mutation did not completely eliminate, Rad51 binding in this assay.

I I L D
Relative ␤-galactosidase activity GFP fusion with wild-type BRC4 binds to Rad51 in cells. Importantly, expression of wild-type, but not the BRC4-M5 mutant, significantly reduced BRCA2 in the Rad51 immunoprecipitates and, in the reciprocal experiment, reduced Rad51 in the BRCA2 immunoprecipitates (Fig. 1D, compare lane 4 with lane 2). These data strongly suggest that conditional expression of a wild-type, but not a mutated, BRC4 repeat effectively disrupts the interaction between BRCA2 and Rad51.
The important role of Rad51 in recombinational DNA doublestrand break repair (13) suggests that disruption of the interaction between BRCA2 and Rad51 may have an adverse effect on the ability of cells to respond to DNA damage. To test this possibility, both WT-8 and MT-11 cells cultured either with or without tetracycline, were mock-exposed or ␥-irradiated (3 Gy), and cell survival was determined by clonogenic assay. Induction of the expression of a wild-type BRC4 repeat (WT-8, ϪTet) significantly reduced cell survival rate when compared with the uninduced (WT-8, ϩ-Tet) or to either the induced (MT-11, ϪTet) or uninduced (MT-11, ϩTet) mutant BRC4 (p Ͻ 0.0001) (Table II).
To further explore DNA damage response phenotypes of these two cell lines, the appearance of radiation-induced Rad51-containing foci was examined. Under uninduced conditions (ϩTet), both clonal lines formed Rad51 foci after ␥-irradiation ( Fig. 2A). However, WT-8 cells induced to express a wild-type GFP-BRC4 repeat exhibited a reduction in the appearance of Rad51 foci compared with MT-11 cells induced to express the GFP-BRC4-M5 mutated repeat ( Fig. 2A for representative field and Fig. 2B for quantification). These data suggest that the interaction between BRCA2 and Rad51 is crucial for the formation of Rad51 repair foci and, furthermore, that exogenous expression of BRC repeats can interfere with this activity.
Increased sensitivity to ionizing radiation may result from defects in the DNA repair machinery or in the molecules essential for cell cycle checkpoint control. When normal mouse embryo fibroblasts are exposed to ␥-irradiation, their transit through the cell cycle is arrested at either one of two points (27,29). The G 1 /S checkpoint, dependent on p53 and p21 (29 -32), prevents the replication of damaged DNA. The G 2 /M checkpoint prevents segregation of damaged chromosomes (33). To test for a potential role of BRCA2-Rad51 interactions in DNA damage-induced cell cycle checkpoint control, cells expressing the GFP-BRC4 repeat were assayed for G 1 /S and G 2 /M checkpoint integrity in response to ␥-irradiation. As shown in Fig.  3A, WT-8 and MT-11 cells, under all conditions, demonstrated nearly identical numbers of BrdUrd-incorporated cells, indicating that the expression of the BRC repeats did not significantly impair G 1 /S checkpoint control in response to ␥-irradiation.
In contrast, when cells were assayed for mitotic figures at variable times after ␥-irradiation, the number of cells in mitosis was not significantly decreased in WT-8 cells induced to express a wild-type BRC4 repeat (Fig. 3B, panel a). However, MT-11 cells induced to express a GFP-BRC4-M5 mutated repeat, as well as uninduced cells, demonstrated significant reductions in mitotic figures (Fig. 3B, panel a). In a parallel experiment, cells were irradiated with 4 -16 Gy. In the population induced to express a wild-type BRC4, the percentage of mitotic cells was significantly higher than that of cells either uninduced or induced to express mutant GFP-BRC4-M5 (Fig.  3B, panel b). These results suggest that cellular expression of a wild-type BRC4 repeat interferes with the radiation-induced G 2 /M checkpoint.
It was reported that mouse embryo fibroblasts, with a genotype Brca2 tm1Cam and predicted to express a C-terminally truncated BRCA2 at amino acid 1492, have intact cell cycle checkpoint responses (8). This truncated BRCA2 protein possesses the first three BRC repeats (34). The abrogation of embryonic lethality by Brca2 tm1Cam (8) compared with other truncating mutations that delete exon 11 of all BRC repeats (7) strongly suggests that the remaining 3 BRC repeats in the  (panels a, c, e, and g) after incubation in the presence (panels a, b, e, and f) or absence (panels c, d, g, and h) of tetracycline (Tet). Fluorescence overlaid with phase-contrast images (panels b, d, f, and h) show nuclear localization of these fusion proteins. C, co-immunoprecipitation of GFP-BRC4 with Rad51 in cells. Cells expressing wild-type (WT-8) or mutated GFP-BRC4 (MT-11) were immunoprecipitated with ␣-myc antibody (top two panels), immunoblotted with either ␣-GFP to detect the GFP-BRC4 fusions or ␣-Rad51 (indicated in the left margin). Immunoprecipitation and Western blotting with ␣-Rad51 antibody (bottom panel) determined the relative levels of endogenous Rad51. D, expression of wild-type GFP-BRC4 in WT-8 cells reduces the complex formation between BRCA2 and Rad51. Thirty-six hours after induction of GFP-BRC4 expression, cell lysates were co-immunoprecipitated with either ␣-BRCA2 (top panel) or ␣-Rad51 antibody (bottom panel). The resulting immune complexes were analyzed by immunoblot analysis with either ␣-BRCA2 or ␣-Rad51 antibody, indicated on the left margin.

TABLE II
Clonogenicity of WT-8 and MT-11 cells after ␥-irradiation Actively growing WT-8 and MT-11 cells (5000) were either induced (ϪTet) or uninduced (ϩTet) to express GFP-BRC4. About 24 h later, cells were either mock-exposed or ␥-irradiated (3 Gy) and cultured for 14 days. Survival rates by colony formation (Ͼ50 cells/colony) were determined by counting the number of colonies per plate. Averages and S.D. were calculated from eight plates. Survival rates were calculated by dividing the number of colonies in the mock-exposed control by the number from exposed cells. Note that the expression of wild-type, but not mutated BRC4 repeat, significantly reduced cell survival in this assay (p Ͻ 0.0001). Brca2 tm1Cam truncated protein are partially functional. The functional importance of the BRC repeats is supported by our results demonstrating that the expression of GFP-BRC4 repeat in cells results in hypersensitivity to radiation and a failure in G 2 /M checkpoint control. The increased radiosensitivity in cells expressing the BRC4 repeat indicates that the complex of BRCA2 and Rad51 is important for the mechanics of the repair process. This notion is compatible with the known role of Rad51 in recombination repair. In this capacity, BRCA2 may facilitate Rad51 function in strand exchange by modulating formation of the Rad51-DNA nucleoprotein filament and/or pairing and strand-exchange steps of DNA double strand break repair (13). The failure of radiation-induced G 2 /M checkpoint control in cells expressing BRC4 repeat indicates that BRCA2 may have a role in this task. It is possible that the formation of the BRCA2 and Rad51 complex could be important for radiationinduced G 2 /M checkpoint control. However, the BRCA2-Rad51 complex is formed independently of DNA damage (10 -12). Other G 2 /M checkpoint proteins may be required to participate in the BRCA2-Rad51 complex. Interestingly, mouse embryo fibroblasts expressing exon 11-deleted Brca1 also exhibit defects in radiation-induced G 2 /M checkpoint control (28) and BRCA2 apparently interacts with BRCA1 (35). It is possible, therefore, that the BRCA2-Rad51 complex may interact with BRCA1 to establish G 2 /M checkpoint control. Alternatively, the BRC repeats may mediate separate interactions with cell cycle checkpoint proteins when induced by DNA damage signaling.
Regardless of which possibility is operative in cells, the data presented here support and extend the model originally proposed (10), that BRC repeat interactions with Rad51 are important for the cellular DNA damage response. The observation that mRad51 Ϫ/Ϫ mouse embryos are not viable and ES cells with the same genotype cannot survive (16,17), coupled with the observation that Brca2 tm1Cam cells are viable but radiationsensitive (8), suggests that Rad51, through interactions with BRCA2, may have functions in addition to DNA repair. Like, BRCA1, it appears that BRCA2 functions in a pathway that bifurcates into one that is important for repair of genetic lesions and another important for restraining cell division until repair is complete. It is possible that each of the BRC repeats, or certain group of the repeats, may have separate and distinct functions within each pathway. Nevertheless, the fact that expression of one BRC repeat can disrupt Rad51-BRCA2 interactions, interfere with the G 2 /M checkpoint, and can make cells radiation-sensitive suggests that it could be used to radiosensitize and/or chemosensitize resistant tumors. Mitotic indices of mock-exposed cells were used as controls. Over 2000 cells were counted at each exposure and time point.