BRCA1 Physically and Functionally Interacts with ATF1*

BRCA1, a breast and ovarian cancer susceptibility gene, encodes a 220-kDa protein whose precise biochemical function remains unclear. BRCA1 contains an N-terminal RING finger that mediates protein-protein interaction. The C-terminal domain of BRCA1 (BRCT) can activate transcription and interacts with RNA polymerase holoenzyme. Using the yeast two-hybrid system, we identified an interaction between the BRCA1 RING finger and ATF1, a member of the cAMP response element-binding protein/activating transcription factor (CREB/ATF) family. We demonstrate that BRCA1 and ATF1 can physically associate in vitro, in yeast, and in human cells. BRCA1 stimulated transcription from a cAMP response element reporter gene in transient transfections. BRCA1 also stimulated transcription from a natural promoter, that of tumor necrosis factor-α, in a manner dependent on the integrity of the cAMP response element. These results implicate BRCA1 in transcriptional activation of ATF1 target genes, some of which are involved in the transcriptional response to DNA damage.

The BRCA1 gene encodes a nuclear protein of 1863 amino acids and fulfills genetic criteria for a tumor suppressor gene in breast and ovarian cancer (1). Mutations in BRCA1 account for a significant proportion of hereditary breast cancer, which represents 5-10% of total breast cancer cases. Although BRCA1 is not mutated in sporadic breast cancer, recent reports suggest that it may be present at reduced levels in high-grade breast carcinoma (2). Although the majority of mutations in BRCA1 result in translation frameshift and premature truncation, several recurring missense mutations were identified. Two of these missense mutations disrupt critical cysteine residues (Cys 61 and Cys 64 ) in the N-terminal RING finger of BRCA1. This observation prompted us to search for proteins capable of interacting with the BRCA1 RING domain.
RING fingers are ϳ60-amino acid-long motifs with a con-served C 3 HC 4 structure that coordinates the binding of two zinc atoms (4). RING motifs are found in Ͼ80 proteins and are thought to mediate protein-protein interactions via a structure that is unique from other zinc fingers (5,6). The BRCA1 RING finger encompasses residues 1-64 and is virtually identical in the human, mouse, and rat BRCA1 homologues (7). BARD1, a RING finger-containing protein, was shown to interact with the BRCA1 RING finger, yet the functional significance of this interaction is not yet understood (8). Missense mutations of cysteine residues in the BRCA1 RING finger indicate that this domain is likely to be vital for the function of the BRCA1 protein (7). The missense mutant C61G is among the most commonly identified mutations in BRCA1 and was shown to disrupt zinc binding and to interfere with homodimerization in vitro (9). Several lines of evidence strongly suggest that BRCA1 is involved in transcription (reviewed in Ref. 10). We and others have observed that the C terminus of BRCA1 activates transcription of a reporter gene when fused to a heterologous DNAbinding domain in yeast and mammalian cells (11)(12)(13). Tumorderived missense mutations in the C terminus are defective for transcriptional activation, which indicates that this may be a relevant function for BRCA1. Furthermore, BRCA1 was shown to interact directly with RNA helicase A, a component of RNA polymerase II holoenzyme, via its C-terminal domain (BRCT) (14,15) and with the p300 coactivator through the C-terminal as well as N-terminal domains (16). We observed that overexpression of wild-type BRCA1, but not tumor-derived mutants, can trigger a G 1 arrest that is mediated by transcriptional activation of p21 Waf1/Cip1 (17). In addition to its association with holoenzyme, BRCA1 can bind to several different transcription factors, including p53, Myc, STAT1, 1 and the CBPinteracting protein CtIP (18 -22).
The spatial distribution and post-translational state of BRCA1 are regulated in response to DNA-damaging agents. BRCA1 was shown to interact with human RAD51, one of several mammalian homologues of bacterial RecA (23). Human RAD51 can mediate ATP-dependent DNA strand exchange reactions in recombination and DNA repair (reviewed in Ref. 24). BRCA1 colocalizes with human RAD51 during S phase in discrete subnuclear foci of mitotic cells and on synaptonemal complexes of cells undergoing meiosis. In response to DNAdamaging agents, including UV, ionizing radiation, and hydroxyurea, BRCA1 subnuclear localization is altered and found to overlap with proliferating cell nuclear antigen-containing nuclear foci at structures undergoing DNA replication (25). These changes correspond with an increase in apparent BRCA1 molecular mass due to phosphorylation. This was recently found to be due to the Chk2 kinase that also regulates p53 activity (26).
The modification of BRCA1 by a cell cycle checkpoint kinase adds to accumulating data that implicate BRCA1 in the maintenance of genome integrity. BRCA1 Ϫ/Ϫ embryonic stem (ES) cells were shown to have an impaired ability to perform transcription-coupled DNA repair (TCR) of oxidative damage induced by ionizing radiation or H 2 O 2 (27). Nullizygous ES cells show a specific defect in TCR. Although BRCA1 was shown to be phosphorylated and relocalized within the nucleus after UV treatment (25), BRCA1 Ϫ/Ϫ ES cells were not impaired in their ability to perform TCR after UV exposure. This suggests that BRCA1 may be directly involved in TCR or that BRCA1 may regulate genes necessary for TCR. Further experiments in BRCA1 Ϫ/Ϫ ES cells and cell lines have supported a role for BRCA1 in DNA repair of double-stranded DNA breaks (28,29) and found that reinsertion of BRCA1 into cell lines can rescue the response to DNA damage (30,31). A complex between BRCA1 and BRCA2, the second breast cancer susceptibility gene product, was identified (32). Evidence suggests that BRCA2 is actively involved in ensuring fidelity of doublestranded DNA break repair (33,34). Taken together, these interactions implicate BRCA1 as part of a macromolecular complex including human RAD51 and BRCA2 that plays a role in DNA repair.
Here we report a specific physical association between the BRCA1 RING domain and ATF1, a member of the CREB/ATF family of basic zipper transcription factors. We demonstrate that BRCA1 is capable of modulating the activity of a reporter gene for CREB/ATF family members. Expression of BRCA1 stimulates the activity of a natural promoter containing a CREB/ATF response element, but fails to stimulate the promoter with a mutated CREB/ATF response element. ATF1 has been implicated as part of a signal transduction pathway that responds to UV damage. These data suggest that BRCA1 is a transcriptional coactivator that can modulate the activity of ATF1. This observation strengthens the hypothesis that BRCA1 has a role in both transcription and DNA damage response.

MATERIALS AND METHODS
Yeast Two-hybrid Screen-The Gal4 Matchmaker II system (CLON-TECH) was used according to the manufacturer's recommendations. Yeast strain CG1945 was cotransformed with pAS2-RING and a human thymus cDNA library (CLONTECH). Transformants were plated on selective medium with 15 mM 3-aminotriazole, which sets a high stringency for interaction. A total of 37 His ϩ /␤-gal ϩ clones were identified, and plasmid DNA was isolated from 23 out of the 37 clones and subjected to DNA sequencing at the Gal4-binding domain junction. Yeast strain CG1945 was retransformed with the indicated combinations of plasmids (see figure legends) according to standard polyethylene glycol/ lithium acetate-mediated transformation. ␤-Galactosidase assays were performed by filter lifts and were terminated after 6 h.
In Vitro Binding Assay-Glutathione S-transferase (GST) fusion proteins were expressed in Escherichia coli strain BL21. Bacteria were grown to mid-log phase, induced with 0.1 mM isopropyl-␤-D-thiogalactopyranoside, grown for 4 h at 30°C, and collected by centrifugation. Bacteria were washed, resuspended in phosphate-buffered saline with 3% Triton X-100, and lysed by sonication, and the supernatant was clarified by centrifugation at 10,000 rpm for 30 min. GST fusion proteins were isolated by incubation of the lysate with 50 l of glutathioneagarose beads (Sigma) for 1 h at 4°C. Beads were collected by centrifugation and washed three times in ice-cold phosphate-buffered saline. An aliquot of beads was boiled in 2ϫ SDS loading buffer, separated by electrophoresis through a 12% polyacrylamide gel, and Coomassiestained to analyze bound proteins.
GST fusion protein (20 g) immobilized on glutathione-agarose beads was used for each in vitro pull-down assay. GST fusion proteins were incubated in 3% bovine serum albumin/NETN buffer (10 mM NaCl, 1 mM EDTA, 20 mM Tris (pH 8.0), 0.2% Nonidet P-40, 1 mM dithiothreitol, and Complete protease inhibitors (Roche Molecular Biochemicals)) to reduce nonspecific binding. Binding assays were conducted in 1 ml of NETN buffer with 30 l of in vitro transcribed/ translated [ 35 S]methionine-labeled protein (Promega). After a 1-h incubation, beads were collected by centrifugation, washed three times in NETN buffer, and boiled in 2ϫ SDS loading buffer. Bound proteins were separated by electrophoresis through a 12% polyacrylamide gel, which was fixed in 30% methanol and 10% acetic acid for 30 min, soaked in water for 30 min, incubated in 1 M sodium salicylate for 30 min, dried, and exposed to film at Ϫ80°C overnight.
Immunoprecipitations and Immunoblotting-For immunoprecipitations, 2 ϫ 10 6 293T cells were seeded 1 day prior to transfection in 10-cm dishes. The cells were transfected with the indicated combination of expression vectors (see figure legends) by calcium phosphate-mediated transfection with 20 g of total DNA. Cells were harvested 48 h post-transfection and lysed in 1 ml of Nonidet P-40 lysis buffer (1% Nonidet P-40, 150 mM NaCl, 5 mM EDTA, and Complete protease inhibitors), and cell lysates were clarified by centrifugation at 18,000 ϫ g for 15 min. BRCA1 was immunoprecipitated with 4 g of affinitypurified anti-BRCA1 polyclonal antibody (Pharmingen catalog no. 66046N) during an overnight incubation with 30 l of a 50% slurry of protein A-agarose (Roche Molecular Biochemicals). Immune complexes were collected by low speed centrifugation, washed three times in 1% Nonidet P-40 lysis buffer, and boiled in 2ϫ SDS loading buffer, and denatured proteins were separated by SDS-polyacrylamide gel electrophoresis. Proteins were transferred to Immobilon (Millipore Corp.), which was blocked in 5% nonfat milk, 150 mM NaCl, 10 mM Tris (pH 8.0), and 0.05% Tween. Immunoblots were performed with sheep anti-ATF1 antisera at 1.5 g/ml (Upstate Biotechnology, Inc.) or AB1, an anti-BRCA1 monoclonal antibody, at 1 g/ml (Calbiochem) and developed by enhanced chemiluminescence (Amersham Pharmacia Biotech).
For immunoprecipitations from MCF-7 cells, subconfluent cells were lysed in 0.25% Nonidet P-40 lysis buffer containing 150 mM NaCl, 50 mM Tris (pH 8.0), and Complete protease inhibitors. Cell lysates were clarified by centrifugation, and immunoprecipitations were performed with 2 g of AB3 (Calbiochem), 2 g of control antibody, or 2 g of AB3 and 2 g of AB1 for 4 h at 4°C with 30 l of a 50% slurry of protein G-agarose (Roche Molecular Biochemicals). Immune complexes were collected by low speed centrifugation, washed three times in 0.25% Nonidet P-40 lysis buffer, boiled in 2ϫ SDS loading buffer, and subjected to electrophoresis. Immunoblotting was performed as described above with 25C10G (Santa Cruz Biotechnology), a murine monoclonal antibody to ATF1, at a concentration of 0.4 g/ml.
Transfection Assays for Transcription-One day prior to transfection, human 293T cells were seeded at a density of 2 ϫ 10 5 cells/well in 12-well dishes. Cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and transfected with Superfect (QIAGEN Inc.) according to the manufacturer's recommendations. Transfections were performed with a constant total of 2 g of DNA, 10 l of Superfect, and 50 ng of tk-Renilla (Promega) and pBluescript (Stratagene). The masses of effector and reporter plasmids are indicated in the figure legends. The CRE-Luc reporter gene was obtained from Stratagene. The TNF-␣ and TNF-␣ mutant CRE reporter plasmids (a gift of J. S. Economou, UCLA) were described previously (35). Forskolin treatments (10 M final concentration; Sigma) were performed for 4 h prior to lysis. Cells were harvested 48 h post-transfection, and 20 l of cell lysate was assayed for luciferase activity according to the manufacturer's recommendations (Dual Luciferase, Promega). For experiments with Gal4 fusion proteins, 100 ng of G5tk-Luc reporter was cotransfected with 50 ng of vectors encoding Gal4 fusion proteins and 50 ng of a cytomegalovirus expression vector containing BRCA1 or empty expression vector.
Plasmid Construction-BRCA1 expression vectors were previously described (17). For immunoprecipitation experiments, a pEF-BOS vector (36) containing full-length human BRCA1 (a gift of Toru Ouchi, Mount Sinai School of Medicine) was used. Full-length human ATF1 was present in clone 31 isolated from the two-hybrid screen described above. ATF1-(1-271) was also amplified by polymerase chain reaction using the N-terminal primer 5Ј-CGGAATTCATGGAAGATTCCCA-CAAG-3Ј and the C-terminal primer 5Ј-TAGTCTAGATTATTTCAATT-GGGGGTCATC-3Ј recloned into pCRII (Invitrogen), which served as the plasmid template for coupled in vitro transcription/translation (Promega). ATF1 deletion mutants were constructed by polymerase chain reaction using the following primers paired with the above primers: ATF1-(73-271), N-terminal primer 5Ј-CCGAATCTCTGAAGATA-CACGGGGC-3Ј; and ATF1-(1-215), C-terminal primer 5Ј-TAGTCTA-GATTAATATTCTTTCTTCTTTCTGAG-3Ј. The resulting products were cloned into the pcDNA3 vector (Invitrogen) for in vitro translation. The BamHI/EcoRI fragment encoding full-length human ATF1 was subsequently recloned into pGEX5X-1 (Amersham Pharmacia Biotech) as a GST fusion construct and into pcDNA3 for expression studies. All constructs were sequenced to verify correct reading frames for fusion genes, and protein expression was verified by immunoblot analysis. The Gal4-(1-147)-ATF1 fusion construct was made by amplifying the ATF1 open reading frame and cloning this fragment into pBXG1, a Gal4 DNA-binding domain mammalian expression vector (37). Gal4-CREB was a gift of Nic Jones (Imperial Cancer Research Fund, London).

ATF1 Identified in Yeast Two-hybrid Screen as a BRCA1-
interacting Protein-In an effort to identify BRCA1 partner proteins, we cloned the N-terminal 101 residues of BRCA1 into pAS2-1 (CLONTECH). The resulting vector, pAS2-RING, was used as a bait to screen a human thymus cDNA library in yeast strain CG1945. One of the 37 His ϩ /␤-gal ϩ clones we identified (clone 31) encoded ATF1, a member of the CREB/ATF family. After plasmid isolation, we retransformed CG1945 with clone 31 alone, with the original bait (pAS2-RING), or with several control plasmids. We observed positive ␤-galactosidase expression only with the combination of clone 31 and pAS2-RING (Table I). We engineered two tumor-derived missense mutations in the RING domain into our bait vector and tested these for interaction with clone 31. Yeast transformed with clone 31 and pAS2-RING-C61G or pAS2-RING-C64G did not show ␤-galactosidase expression (Table I), indicating that clone 31 produces a protein capable of an interaction with wild-type RING but not mutant RING domains.
Sequencing of clone 31 surprisingly indicated the presence of sequences encoding the Gal4 activation domain, a few amino acids, and a stop codon. This was followed by the 5Ј-untranslated region of ATF1, the start codon, and the entire open reading frame for ATF1. We hypothesized that clone 31 directed the expression of ATF1 from a cryptic promoter and that this expression was responsible for the positive interaction we observed. Efforts to detect the expression of ATF1 in yeast from the original plasmid (clone 31) were hampered by poor specificity with two commercial anti-ATF1 antibodies. We note that other groups have identified CREB from two-hybrid screens as a cDNA insert both out of frame and backwards with respect to the Gal4 activation domain (38). Therefore, to determine whether ATF1 could interact with the BRCA1 RING domain in yeast, we recloned ATF1 into pGAD424 as an in-frame fusion protein with the Gal4 activation domain. This plasmid was sequenced to verify the correct reading frame and was used to test for a specific interaction with pAS2-RING. This plasmid, designated pGAD-ATF1, was used to test for a specific interaction with pAS2-RING in subsequent experiments.
The combination of pAS2-RING with pGAD-ATF1 yielded expression of the ␤-galactosidase reporter gene in yeast (Fig.  1A). Neither plasmid alone was sufficient to activate the reporter gene. We again tested the ability of the two missense mutations in the RING domain (C61G and C64G) to interact with pGAD-ATF1. Yeast transformed with either pAS2-RING-C61G or pAS2-RING-C64G and pGAD-ATF1 show marked reduction in the expression of ␤-galactosidase (Fig. 1A) despite comparable levels of expression of wild-type and mutant Gal4-BRCA1 fusion proteins (Fig. 1B). This was identical to the pattern observed for the original clone 31. Thus, we observed a specific interaction between full-length ATF1 and the BRCA1 RING domain in the yeast two-hybrid system. In yeast, the BRCA1 RING domain was sufficient to promote an interaction with ATF1, which was abrogated by either of the two tumorderived mutations, C61G or C64G.
We confirmed the interaction between the BRCA1 RING domain and ATF1 using a GST pull-down assay. GST fusion proteins containing BRCA1 residues 1-101 or 1-304 were constructed, expressed in E. coli, and purified on glutathioneagarose beads. GST-BRCA1 fusion proteins captured in vitro transcribed/translated [ 35 S]methionine-labeled ATF1 ( Fig. 2A). GST alone did not capture radiolabeled ATF1. Lower molecular mass species may be degradation products or failure products of the in vitro translation. We performed the reciprocal experiment with affinity-purified GST-ATF1. GST-ATF1 bound to glutathione-agarose beads was incubated with radiolabeled Nterminal BRCA1 (residues 1-304). GST-ATF1 captured radiolabeled BRCA1 to a greater extent than GST alone, yet the interaction appeared to have a relatively low affinity in vitro (6-fold increase above GST alone by densitometric analysis) (Fig. 2B). To localize the region of interaction between the two proteins, we made two deletion mutants of ATF1. The first mutant, ATF1-(73-271), lacks the first 72 amino acids of ATF and therefore the activation and CBP-binding domains. The second mutant, ATF1-(1-215), lacks the C-terminal basic zipper DNA-binding region. ATF1-(1-215) did not demonstrate appreciable binding to GST-BRCA1-(1-304), whereas ATF1-(73-271) still bound to the BRCA1 affinity reagent (Fig. 2C). This indicates that the BRCA1-ATF1 interaction is specific and is dependent on the RING finger of BRCA1 and the DNAbinding domain of ATF1.
Full-length BRCA1 Interacts with ATF1 in Mammalian Cells-To determine whether BRCA1 could interact with ATF1 in human cells, we expressed full-length BRCA1 in human 293T cells and immunoprecipitated a 220-kDa species with rabbit polyclonal antisera. ATF1 coprecipitated with BRCA1 from cells cotransfected with both BRCA1 and ATF1 expression vectors (Fig. 3A). In 293T cell lysates, the coprecipitation of ATF1 with BRCA1 occurred only when both proteins were overexpressed (Fig. 3A). The most important indication of the physiological significance of the BRCA1-ATF1 interaction was the demonstration of an endogenous complex between BRCA1 and ATF1. Lysates from rapidly growing human MCF-7 breast cancer cells were subjected to immunoprecipitation with a monoclonal antibody directed against the C terminus of BRCA1 (AB3), followed by immunoblotting with an anti-ATF1 antibody. ATF1 was identified by immunoblot analysis as a 38-kDa species present in BRCA1 immunoprecipitates (Fig. 3B). We also observed that AB1, a monoclonal antibody directed against the N terminus of BRCA1, was incapable of coprecipitating  31 shows a positive interaction with the wild-type but not mutant BRCA1 RING finger pAS2-1 contains the Gal4 DNA-binding domain. pAS2-RING encodes residues 1-101 of BRCA1 fused in frame to the Gal4 DNA-binding domain. pAS2-RING-C61G and pAS2-RING-C64G contain missense mutations in the RING finger as described under "Results." pLAM5Ј (Clontech) contains lamin fused to the Gal4 DNA-binding domain and is a routine test for false positives. pVA3 and pTD1 (Clontech) encode p53 and SV40 T antigen, respectively, which are positive controls for protein-protein interaction. ␤-Galactosidase expression was assessed by filter assay using 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (Xgal, Sigma) as a substrate. Ϫ, no detectable blue color; ϩϩ, blue color that developed within 1 h; ϩϩϩ, blue color that developed within 30 min.

Bait vector
Prey vector ␤-Galactosidase expression pAS2-1 pTD1 ϩϩϩ ATF1 and in fact might have blocked the interaction between BRCA1 and ATF1. Immunoprecipitation with the combination of AB1 and AB3 (Fig. 3B, third lane) resulted in a reduction of coprecipitating ATF1, supporting the notion that an anti-BRCA1 monoclonal antibody directed against the RING finger can block the interaction between BRCA1 and ATF1. BRCA1 Modulates Transactivation of Gal4-ATF1 and Gal4-CREB Fusion Proteins-Because BRCA1 was shown to regulate transcription (17) and the activity of specific transcription factors (18 -21), we sought to determine whether BRCA1 could modulate the transcriptional activity of ATF1 or CREB. We constructed a Gal4-ATF1 fusion protein, which activated a luciferase reporter gene with Gal4 DNA-binding sites (G5-Luc) in the presence of forskolin, an adenylate cyclase agonist, and activator of protein kinase A. Expression of Gal4-ATF1 or Gal4-CREB in the absence of forskolin had a minimal (Ͻ1.5-fold) effect on the activity of the reporter. BRCA1 alone had no effect on the activity of the G5-Luc reporter. 2 Transfection of BRCA1 in combination with Gal4-ATF1 in the presence of forskolin activated transcription of the Gal4-dependent reporter by ϳ3fold (Fig. 4). Full-length BRCA1 containing the missense mutation C64G was markedly defective in its ability to modulate Gal4-ATF1 transactivation; however, the missense mutant C61G had activity similar to wild-type BRCA1 in stimulating Gal4-ATF1 transactivation. Thus, expression of BRCA1 stim-ulates the activity of Gal4-ATF1, and this effect is disrupted by one of two tumor-derived RING missense mutations. BRCA1 also augmented transcriptional activation by a Gal4-CREB fusion protein on the G5-Luc reporter. As shown in Fig. 4, wild-type BRCA1 elicited a 3-fold increase in Gal4-CREB transcriptional activation in the presence of forskolin. As with Gal4-ATF1, BRCA1 C64G, but not C61G, was significantly defective in augmenting Gal4-CREB transactivation of the Gal4-dependent luciferase reporter gene.
BRCA1 Modulates Transactivation through the cAMP Response Element-To assess whether BRCA1 is involved in the functional activation of CREB/ATF transcription factors, we tested the ability of BRCA1 to stimulate transcription from a CRE reporter gene. The activity of the CRE-Luc reporter plasmid measures activation by CREB/ATF family members endogenously present in 293T cells. This reporter gene was minimally active in the absence of forskolin, an adenylate cyclase agonist. Expression of BRCA1 alone, in the absence of exogenous forskolin, did not result in activation of the CRE-Luc reporter gene (Fig. 5B). Expression of BRCA1 in the presence of forskolin stimulated the CRE-Luc reporter gene 3-fold (Fig.  5A). We next tested the ability of mutant BRCA1 species to augment CRE-Luc reporter activity. The tumor-derived mutants C61G and C64G both showed differential activation compared with wild-type BRCA1 when expressed in the context of the full-length protein. BRCA1 C61G showed an ϳ40% reduction in its ability to enhance transactivation from the CRE-Luc reporter (Fig. 5A). BRCA1 C64G was completely inactive in augmenting CRE-Luc transactivation (Fig. 5A). Both mutant BRCA1 species were produced at levels comparable to wildtype BRCA1 as determined by immunoblot analysis (Fig. 5C).
To determine whether BRCA1 could affect the transcription of a natural promoter containing a CRE, we tested the effect of BRCA1 on the TNF-␣ promoter. TNF-␣ is a major extracellular signal for programmed cell death. Its expression is responsive to DNA damage, potentially mediating cell survival (39). The TNF-␣ promoter was transfected into 293T cells in the presence or absence of BRCA1. BRCA1 stimulated expression of this promoter 2-3-fold, consistent with our result with the model CRE-Luc promoter (Fig. 6). A mutant form of the TNF-␣ promoter containing a single mutation in the CRE site had a much lower basal activity compared with the wild-type promoter. BRCA1 could not activate transcription from this promoter. This indicates that transcriptional activation of this natural promoter by BRCA1 is dependent on the integrity of the CRE.

DISCUSSION
The data presented here demonstrate that BRCA1 physically and functionally interacts with ATF1. The N-terminal 101 residues of BRCA1 were sufficient for promoting an interaction between the BRCA1 RING finger and ATF1. In yeast, this interaction was disrupted by either of the two tumor-derived RING missense mutants, C61G or C64G. ATF1 was originally identified as a nuclear protein bound to the human T-cell leukemia virus type 1 long terminal repeat (40,41). ATF1 is highly homologous to CREB, diverging in sequence at the N terminus and virtually identical at the C terminus. Both ATF1 and CREB bind a consensus CRE (TGACGTCA) and mediate transcriptional activation in response to cAMP and Ca 2ϩ (42). ATF1 binds DNA as a homodimer or as a CREB/ATF1 het-erodimer (43). Both CREB and ATF1 share a conserved protein kinase A phosphorylation consensus site (Ser 63 and Ser 133 , respectively) that can also undergo phosphorylation by calcium/calmodulin-dependent kinase I, II, or IV. Phosphorylation of CREB/ATF1 facilitates docking of the transcriptional coactivators and CBP and p300 through interactions between the kinase-inducible domain of CREB/ATF1 and the KIX domain of CBP or p300 (44 -46). ATF1 is involved in a chromosomal translocation (t12;22) in malignant melanoma of soft parts with the Ewing's sarcoma oncogene (EWS) (47). The EWS-ATF1 chimeric protein was shown to act as a constitutive transcriptional activator, which suggests that ATF1 target genes may be deregulated in oncogenesis (48). It is possible that ATF1 may be targeted for mutation in a subset of breast or ovarian cancer cases, as is the case for another BRCA1 RING partner protein, BARD1 (49).
The interaction between the RING finger of BRCA1 and ATF1 was confirmed using a GST pull-down assay using both BRCA1 and ATF1 capture reagents and by co-immunoprecipitation of the endogenous proteins in MCF-7 breast cancer cells. Furthermore, the interaction was dependent on discrete regions of the two proteins, the RING finger motif of BRCA1 and the basic zipper DNA-binding domain of ATF1. The interaction between BRCA1 and ATF1 may be direct, although it is possible that a factor present in the reticulocyte lysate mediates the interaction. Since the C and N termini of BRCA1 can interact with CBP (50) and p300 (16), the interaction between fulllength endogenous BRCA1 and ATF1 may be stabilized by these coactivators.
Expression of BRCA1 increased luciferase activity when the Gal4-ATF1 or Gal4-CREB fusion proteins were used as transcriptional activators on a luciferase reporter containing Gal4 DNA-binding sites. Expression of BRCA1 also stimulated transcription from a CRE-dependent reporter, which measures activity of endogenous CREB/ATF family members, in the presence of the protein kinase A agonist forskolin. The C-terminal FIG. 3. In vivo association of BRCA1 with ATF1. A, 293T cells were transfected with the indicated plasmids as described under "Materials and Methods." Cell lysates were subjected to immunoprecipitation (IP) with anti-BRCA1 polyclonal antisera (Upstate Biotechnology, Inc.). After extensive washing, the immunoprecipitates were separated by SDS-polyacrylamide gel electrophoresis and transferred to a nylon membrane. The upper panel is an immunoblot probed with sheep anti-ATF1 polyclonal antisera. The lower panel is an immunoblot probed with AB1, a mouse anti-BRCA1 monoclonal antibody. Control lysates represent a fraction (5%) of total cellular lysates. B, MCF-7 cells were lysed and subjected to immunoprecipitation with the indicated antibody as described under "Materials and Methods." Anti-FLAG antibody (M2, Sigma) was used as a control (Con) for immunoprecipitation. The immunoblot was probed with 25C10G, a murine anti-ATF1 monoclonal antibody, and developed by enhanced chemiluminescence. DNA-binding domain of ATF1 was required for in vitro interaction with BRCA1 and not the N-terminal KIX domain, which is phosphorylated by protein kinase A and mediates interaction of CREB/ATF family members with CBP. We also found no difference in the interaction between endogenous BRCA1 and ATF1 in the presence or absence of forskolin in MCF-7 cells. 2 These data suggest a model in which BRCA1 binds to the C terminus of ATF1, but cannot modulate the transcriptional activity of the protein unless ATF1 engages the CBP/p300 coactivators through its N-terminal domain.
BRCA1 stimulated the expression of a natural promoter containing a CRE, the TNF-␣ promoter. CREB and other ATF family members have been shown to be capable of binding to the TNF-␣ CRE (51). Expression of BRCA1 stimulated the wild-type TNF-␣ promoter, but failed to stimulate the mutant CRE promoter. These data suggest that BRCA1 can function as a transcriptional coactivator for CREB/ATF family members. Regulation of the TNF-␣ promoter is controlled by AP1, EGR1, CAAT/enhancer-binding protein-␤, and CREB/ATF transcription factors (35,(52)(53)(54). Mutation of the TNF-␣ CRE resulted in a Ͼ10-fold decrease in promoter activity, but some residual activity was still present. Intriguingly, TNF-␣ expression has been linked to hypoxia and both ionizing and UV irradiation in mammalian cells (51,55,56). Thus, it is possible that BRCA1 cooperates with CREB/ATF family members to regulate TNF-␣ in response to DNA-damaging agents.
Several CREB/ATF target genes are directly connected with repair of damaged DNA. In HeLa cells treated with the DNA- alkylating agent N-methyl-N-nitro-N-nitrosoguanidine, DNA polymerase ␤, the polymerase involved in nucleotide base excision repair, is up-regulated (57). The transcriptional induction of polymerase ␤ is mediated through a CRE, and treatment with N-methyl-N-nitro-N-nitrosoguanidine triggers CREB phosphorylation. Similarly, UV-C exposure triggers the transcriptional activation of c-Fos through a CRE in its promoter (58). CREB and ATF1 undergo rapid phosphorylation in response to UV-C (59) and may direct the transcriptional activation of target genes involved in DNA damage response. BRCA1 may modulate and augment this activity.
Two viewpoints regarding the function of BRCA1 have emerged in recent years. Substantial evidence implicates BRCA1 in transcription, including association with RNA polymerase holoenzyme (10,60). On the other hand, BRCA1 association with human RAD51 and BRCA2 and the identification of BRCA1 as a component of a supercomplex of proteins involved in sensing DNA damage and transcription-coupled repair suggest that BRCA1 is critical in the maintenance of genome integrity (reviewed in Ref. 61). These two models of BRCA1 function are not mutually exclusive. Both BRCA1 and ATF1 are phosphorylated in response to DNA damage; and in the case of ATF1, phosphorylation activates docking with coactivators and transcriptional activation. BRCA1 itself may be under the transcriptional control of CREB/ATF family members since a recent report showed that a CRE in the BRCA1 proximal promoter is methylated in breast cancer, correlating with decreased BRCA1 expression (62). The interaction between BRCA1 and ATF1 represents a connection between transcriptional activation and DNA damage response. A critical next step will be to identify genes targeted by ATF1 and BRCA1 in response to DNA damage. Intriguingly, proliferating cell nuclear antigen, a gene whose promoter contains a CRE, is up-regulated after DNA damage and is up-regulated after adenovirus-mediated transfer of BRCA1 into a cancer cell line (3). Whether the BRCA1-ATF1 interaction may play a role in this is under investigation.