Binding of CtIP to the BRCT Repeats of BRCA1 Involved in the Transcription Regulation of p21 Is Disrupted Upon DNA Damage*

Mutations in BRCA1 are responsible for nearly all of the hereditary ovarian and breast cancers, and about half of those in breast cancer-only kindreds. The ability of BRCA1 to transactivate the p21 promoter can be inactivated by mutation of the conserved BRCA1 C-terminal (BRCT) repeats. To explore the mechanisms of this BRCA1 function, the BRCT repeats were used as bait in a yeast two-hybrid screen. A known protein, CtIP, a co-repressor with CtBP, was found. CtIP interacts specifically with the BRCT repeats of BRCA1, bothin vitro and in vivo, and tumor-derived mutations in this region abolished these interactions. The association of BRCA1 with CtIP was also abrogated in cells treated with DNA-damaging agents including UV, γ-irradiation, and adriamycin, a response correlated with BRCA1 phosphorylation. The transactivation of the p21 promoter by BRCA1 was diminished by expression of exogenous CtIP and CtBP. These results suggest that the binding of the BRCT repeats of BRCA1 to CtIP/CtBP is critical in mediating transcriptional regulation of p21 in response to DNA damage.

Mutations in BRCA1 are responsible for nearly all of the hereditary ovarian and breast cancers, and about half of those in breast cancer-only kindreds (1)(2)(3). How BRCA1 inactivation leads to tumor formation remains unclear. Studies of homozygous mutation of Brca1 in mice showed a phenotype of early embryonic lethality (4 -6). Interestingly, Brca1 Ϫ/Ϫ mouse embryonic stem cells are hypersensitive to ionizing radiation and hydrogen peroxide, and defective in transcription-coupled repair of oxidative DNA damage (7). An extension of development to embryonic day 11-12 was observed in Brca1 Ϫ/Ϫ mice carrying additional p53 or p21 waf/cip1 null mutations (8,9). A role for BRCA1 in transcription regulation was provided by observations showing that expression of p21, a known target for p53 transcriptional activation, is increased significantly in Brca1 mutant embryos (8). Consistent with this observation, wildtype, but not mutant BRCA1, was able to transactivate the expression of p21 and inhibit cell cycle progression from G 1 into S phase in human cells (10). Taken together, these results suggested that BRCA1 may have a role in the DNA repair process involving p21 and p53 expression. However, the molecular basis for these observations is largely unknown.
The C terminus of BRCA1 contains a transcription activation region (10 -13) and two conserved BRCT 1 repeats frequently found in proteins involved in DNA repair and cell cycle regulation (14 -16). Although the general function of the BRCT motif is unclear, several lines of evidence suggest that it may be involved in protein-protein interactions (17,18). Here, we report specific interactions between CtIP and the BRCT repeats of BRCA1 both in vitro and in vivo. Apparently, complex formation of BRCA1, CtIP and CtBP plays an important role in the regulation of p21 expression.
Yeast Two-hybrid Screen-The pAS-BRCT plasmid, which contains the GAL4 DNA-binding domain fused to BRCA1 (amino acids 1634 -1863), was used as the bait for screening a cDNA library prepared from human B lymphocytes as described (24). Because the bait has weak * This work was supported by Grants from the NCI (to W. H. L.) (CA58183 and CA30195) and from the Susan G. Komen Foundation for Breast Cancer Research (to P. L. C.) (9733). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In Vitro Binding Assay-Bacterially expressed and purified GST or GST fusion proteins were incubated with in vitro synthesized [ 35 S]methionine-labeled CtIP, BRCA1, or CtIP and BRCA1 proteins as described (25).
Mammalian Two-hybrid Assay-Human kidney 293 cells were used in this assay as described (20). The expression of GAL4-X fusion proteins was verified by straight Westerns using a specific antibody that recognizes the GAL4 DNA-binding domain (Santa Cruz Biotechnology, Inc. Santa Cruz, CA).
Immunoprecipitation and Western Blot Analyses-Cells were lysed in lysis 250 buffer and immunoprecipitated as described (24). The immunoprecipitates were separated by SDS-polyacrylamide gel electrophoresis and analyzed by immunoblotting.
Cell Transfection and Luciferase Assay-Human kidney 293 cells were transfected with 0.5 g of pWWW-luc or pG5E-luc, 0.5 g of pSV40-␤-gal, and 2 g of each different plasmid DNA as indicated (control vector pcDNA3.1 was used to bring the final amount of DNA to 10 g) using calcium phosphate/DNA co-precipitation method. Luciferase activity was measured 48 h after transfection as described (26). For each 10-cm dish of HCT116 cells, 10 g of pRcCMV-CtBP plasmid was transfected, using the Lipofectin transfection method.

RESULTS
CtIP Interacts Specifically with Wild-type BRCT Repeats of BRCA1-To explore the potential function of the BRCT repeats of BRCA1, we used them as bait in a yeast two-hybrid screen for interacting proteins (24). One of the 20 clones isolated encodes amino acids 17 to 713 of the known protein CtIP (Fig.  1A), which was also identified by others using a different approach (27,28).
To test whether CtIP directly binds to the BRCT repeats, an in vitro binding assay using GST-fusion proteins was performed. The BRCT repeats that served as the bait in the above screen and a mutant containing a familial 1853 tyrosine nonsense mutation were fused with GST (GST-BRCT and GST-BRCT⌬, Fig. 1B). Bacterially expressed and purified GST-BRCT, but not GST-BRCT⌬ or GST (Fig. 1C), can bind to in vitro synthesized [ 35 S]methionine-labeled CtIP protein (Fig.  1D, lane 3, compare lanes 2 and 4).
Identification of Cellular CtIP Protein-To study the in vivo interactions of endogenous BRCA1 and CtIP, mouse polyclonal antibodies recognizing CtIP (C21) were generated using an antigen consisting of GST translationally fused with amino acids 324 -537 of CtIP. [ 35 S]methionine-labeled human colon carcinoma cell (HCT116) lysates immunoprecipitated with C21 identified a band (Fig. 2iA, lane 2) whose mobility was consistent with the predicted molecular mass (125 kDa) of CtIP (21). A band of similar size was not detected in the immunoprecipitates of pre-immune serum (Fig. 2A, compare lanes 1 and 2). Pre-incubation of the antiserum with GST-CtIP, but not GST resulted in specific depletion of the 125-kDa band (Fig. 2A,  compare lanes 3 and 4). Reprecipitation of the immunoprecipitates by a second incubation with anti-CtIP antibodies resulted in a specific 125-kDa band (Fig. 2A, lane 5). Based on these results, we concluded that the 125-kDa protein is the cellular CtIP protein.
Co-immunoprecipitation of CtIP and BRCA1 in Vivo-The in vivo interaction between CtIP and BRCA1 was further examined in HCT116 and breast cancer cells, HCC1937 (29), by co-immunoprecipitation. HCC1937 cells contain an insertion of cytosine at nucleotide 5382 of BRCA1 that generates a frameshift at amino acid 1794, which stops translation at 1829 (29). Anti-BRCA1 monoclonal antibody 6B4 (30), but not control antibody against p84 (N5), a nuclear matrix protein (31), specifically immunoprecipitated a 220-kDa protein in HCT116 cell lysates (Fig. 2B, compare lanes 1 and 2). Consistent with the 5382insC mutation in BRCA1, a faster migrating product was detected in HCC1937 cell lysates (Fig. 2B, lane 3). This protein is likely to be the HCC1937 BRCA1 product because it cannot be recognized by C-20 antibodies (against amino acids 1843-1862 of BRCA1), which readily immunoprecipitated a 220-kDa protein from HCT116 (Fig. 2B, compare lanes 5 and 4). Furthermore, using anti-BRCA1 6B4, but not control antibody N5, CtIP was co-immunoprecipitated with BRCA1 in HCT116 cell lysates (Fig. 2B, compare lanes 1Ј and 2Ј). No detectable CtIP was present in the 6B4 immunoprecipitates from HCC1937 cell lysate (Fig. 2B, lane 3Ј). Consistent with this result, BRCA1 was reciprocally co-immunoprecipitated with anti-CtIP C21 antibody from HCT116 cell lysate, but not from HCC1937 cell lysate (Fig. 2B, compare lanes 7 and 8). Levels of CtIP immunoprecipitates detected from both cell lines were comparable (Fig. 2B, lanes 7Ј and 8Ј). These data indicated the existence of an in vivo complex of BRCA1 and CtIP, that, in HCC1937 cells, is disrupted presumably by an altered C terminus lacking intact BRCT repeats.
Dissociation of CtIP from BRCA1 upon Treatment with DNAdamaging Agents-Since BRCA1 has a potential role in DNA repair, its interaction with CtIP may mediate cellular responses to DNA damage. Therefore, it is possible that their interaction might be altered upon treatment of cells with DNAdamaging agents. To address this possibility, HCT116 cells were treated with UV, ␥-rays, or adriamycin, and the cell lysates were immunoprecipitated with anti-BRCA1 antibody (6B4). Consistent with previous data (32,33), a slower migrating form of phosphorylated BRCA1 was detected after treatment of the cells with DNA-damaging agents (Fig. 3A, lanes  3-5 compare with lane 2). Importantly, the association of CtIP was undetectable in the BRCA1 immunoprecipitates subse-  4). Lane 5 shows a double immunoprecipitation with anti-CtIP antibody. The anti-CtIP antibody specifically recognizes a cellular protein (arrow) whose mobility is consistent with a molecular mass of 125 kDa. B, co-immunoprecipitation of CtIP with wild-type BRCA1, but not with mutated BRCA1 in vivo. Lysates from HCT116 (lanes 1, 2, 4,  6, and 7) and HCC1937 cells (lanes 3, 5, and 8) were immunoprecipitated with the following antibodies; anti-p84 control monoclonal antibody (lane 1), 6B4, an anti-BRCA1 monoclonal antibody (lanes 2 and 3), C20, a rabbit polyclonal antibody against BRCA1 C-terminal peptide (Santa Cruz Biotechnology, Inc.) (lanes 4 and 5), pre-immune serum (lane 6), and C21, anti-CtIP mouse polyclonal serum (lanes 7 and 8). In the upper panels (lanes 1-8), the blots of immunoprecipitated proteins were probed with anti-BRCA1 monoclonal antibody to detect BRCA1; in the lower panels (lanes 1Ј-3Ј and 6Ј-8Ј), the blots were probed with anti-CtIP polyclonal serum to detect CtIP. CtIP can be co-immunoprecipitated with the wild-type (lanes 2 and 2Ј), but not mutated (lanes 3 and 3Ј) BRCA1. Conversely, wild-type (lanes 7 and 7Ј), but not mutated (lanes 8 and 8Ј) BRCA1 can be co-immunoprecipitated with CtIP. quent to DNA damage (Fig. 3A, compare lanes 3Ј-5Ј and lane  2Ј). This change in BRCA1/CtIP association appeared to be correlated with increased BRCA1 phosphorylation. The undetectable levels of CtIP in BRCA1 immunoprecipitates were not due to lower expression because the amount of CtIP did not change subsequent to treatment with genotoxic agents (Fig.  3B). Similar results were also observed in breast epithelial cell line MCF10A (data not shown).
Association of CtBP and BRCA1 Mediated by CtIP-CtIP was originally identified in a yeast two-hybrid screen for proteins that interacted with an adenovirus E1A C-terminal-binding protein, CtBP (22). The binding of CtBP to E1A represses CR1-dependent transcriptional activation and tumorigenesis (21,34). Both CtIP and E1A use a conserved PLDLS motif to interact with CtBP. Several transcription factors including Knirps, Snail (35), and Hairy (36), which contain P-DLS-K/V motifs, bind to CtBP to repress transcription during Drosophila development. Based on this observation, it is possible that BRCA1 is linked to the CtBP co-repressor through CtIP.
To address this possibility, we used an in vitro binding assay to test whether CtIP has different binding motifs for BRCA1 and CtBP. As shown, mutation of the PLDLS motif in CtIP to LASQC abolished the interaction between CtIP and GST-CtBP (Fig. 4A, compare lanes 4 and 8). However, the interaction between the mutated CtIP and GST-BRCT was unaffected (Fig.  4A, compare lanes 3 and 7), suggesting that CtIP binds to CtBP and the BRCT repeats of BRCA1 using different motifs. Thus, it is possible that CtIP can bridge BRCA1 and CtBP to form a complex. To test this possibility, HCT116 cells were transiently transfected with pRcCMV-CtBP to overexpress the T7-tagged CtBP full-length protein. As shown, anti-T7-tag antibody can co-immunoprecipitate the full-length CtBP protein, cellular CtIP, and BRCA1 (Fig. 4B, lane 2), but not the control antibody (lane 1). In the reciprocal experiment, anti-BRCA1 antibody can bring down cellular BRCA1, CtIP and T7-tagged CtBP (Fig.  4B, lane 4). These results suggest that CtBP, CtIP, and BRCA1 can form a complex.
Repression of BRCA1-dependent Transactivation of the p21 Promoter by CtIP and CtBP-Previous studies (10,12) showed that BRCA1 was able to transactivate p21 expression. Formation of the CtBP, CtIP, and BRCA1 complex predicts that ectopic expression of CtIP or CtBP may affect BRCA1-dependent transactivation of the p21 promoter. To test this hypothesis, transient transfections of human 293 cells were performed with a p21 promoter-luciferase reporter plasmid pWWW-luc, pSV40-␤-gal transfection control plasmid, and combinations of BRCA1, CtIP, and CtBP expression vectors. The expression of BRCA1 resulted in a 5-fold activation of the p21 promoter compared with empty vector alone. Co-expression of CtIP moderately inhibited BRCA1-dependent transactivation of the p21 promoter. However, co-expression of CtIP and CtBP repressed p21 promoter activity to background levels (Fig. 5A). Interestingly, co-expression of CtBP and BRCA1 also resulted in significant repression of p21 promoter, which was likely due to the abundance of endogenous CtIP in cells. Likewise, the modest inhibition of BRCA1-dependent transcription by CtIP may be due to the recruitment of cellular CtBP. Overexpression of CtBP alone does not have an effect on p21 promoter (Fig. 5A). The repression of BRCA1-dependent transactivation of p21 promoter by CtBP and CtIP depends on their association with BRCA1, because overexpression of CtIP and CtBP cannot repress the transactivation of GAL4 promoter (pG5E-Luc) by the GAL4-VP16 hybrid (Fig. 5B). These experiments suggest that a potential biological function of the BRCA1/CtIP interaction is to repress target promoters through contacts with the CtBP co-repressor. DISCUSSION The BRCT repeats were first identified as a highly conserved structural domain among more than 50 nonorthologous proteins, many of which are involved in DNA repair and cell cycle check point control (14 -16). Familial mutations have been frequently found in the BRCT repeats of BRCA1, suggesting that the function of the BRCT repeats of BRCA1 is important  3 and 7), or GST-CtBP (lanes 4 and  8). GST-CtBP binds to wild-type, but not the mutated CtIP (compare lanes 4 and 8). B, HCT116 cells were transfected with pRcCMV-CtBP plasmid. The cell lyates were immunoprecipitated with 8G11 (anti-GST monoclonal antibody, lanes 1 and 3), ␣-T7 (anti-T7-tag monoclonal antibody, lane 2), or 6B4 (anti-BRCA1 monoclonal antibody, lane 4). The membranes were probed with 6B4, C21, or ␣-T7 antibody as indicated. T7-tagged CtBP full-length protein is indicated by an arrow. The IgG heavy chain is marked by an asterisk.
for the BRCA1 tumor suppression function. Using BRCT repeats as bait, we have identified a BRCA1-associated protein, CtIP. CtIP interacts specifically with the BRCT repeats of BRCA1 both in vitro and in vivo, and tumor-derived mutations in BRCT repeats abolish this specific interaction. The three tumorderived mutations A1708E, P1749R, and Y1853term, are either located in the first motif, second motif, or the spacer region of BRCT repeats. However, all of them abolish the interaction of BRCT repeats with CtIP. Interestingly, these mutations also abolish the transactivation activity of BRCA1 (13). This tight correlation suggests that the interaction between BRCA1 and CtIP is relevant to its transcription regulation activity.
CtIP was originally identified as a CtBP-associated protein in a yeast two-hybrid screen. Further study has mapped the PLDLS motif in CtIP as the binding site for CtBP (22). This motif was also found in E1A and several Drosophila transcriptional factors (22,35,36). BRCA1 does not have a PLDLS motif, and cannot bind to CtBP directly. Apparently, CtIP has different binding motifs for CtBP and BRCA1 respectively, allowing CtBP to associate with BRCA1 through CtIP, as shown in Fig.  4. The formation of this complex is important for the repression of BRCA1-dependent transcription activation of the p21 promoter. Data from previous studies (10) and in this report (Fig.  5) suggested that overexpression of exogenous BRCA1 can transactivate the p21 promoter. A reasonable explanation for these observations is that overexpression of BRCA1 titrates the CtIP/CtBP complex and allows the additional copies of BRCA1 to act on the p21 promoter.
The up-regulation of p21 was observed in cells treated with DNA-damaging agents (37). Similarly, treatment with DNAdamaging agents disrupts the interaction between BRCA1 and CtIP/CtBP, thus, allowing BRCA1 to transactivate p21 promoter. Dissociation of CtIP and BRCA1 is correlated with the hyperphosphorylation of BRCA1 upon treatment with DNAdamaging agents. These results suggest that CtIP/CtBP may negatively regulate the transactivation activity of BRCA1 on the p21 promoter. However, whether BRCA1 transactivates p21 promoter directly or binds to additional transcription factors remains to be explored. FIG. 5. CtIP/CtBP specifically represses BRCA1-dependent transactivation of p21 promoter. A, human 293 cells were transfected with pWWW-luc (10), pSV40-␤-gal, and the expression plasmids containing BRCA1, CtIP, and CtBP as indicated. B, similar transfection was performed as shown in panel A, with pG5E-Luc, pSV40-␤-gal and the expression plasmids containing GAL4-VP16, CtIP, and CtBP as indicated. In both experiments luciferase activity was measured 48 h after transfection and normalized to ␤-galactosidase activity. Expression of CtIP and CtBP repress the BRCA1-dependent transactivation of p21 promoter, but not the GAL4-VP16-dependent transactivation of GAL4 promoter. The data represent three independent transfections, each assayed for luciferase activity twice.