RAD51 Interacts with the Evolutionarily Conserved BRC Motifs in the Human Breast Cancer Susceptibility Gene brca2 *

Recent work has shown that the murine BRCA2 tumor suppressor protein interacts with the murine RAD51 protein. This interaction suggests that BRCA2 participates in DNA repair. Residues 3196–3232 of the murine BRCA2 protein were shown to be involved in this interaction. Here, we report the detailed mapping of additional domains that are involved in interactions between the human homologs of these two proteins. Through yeast two-hybrid and biochemical assays, we demonstrate that the RAD51 protein interacts specifically with the eight evolutionarily conserved BRC motifs encoded in exon 11 ofbrca2 and with a similar motif found in aCaenorhabditis elegans hypothetical protein. Deletion analysis demonstrates that residues 98–339 of human RAD51 interact with the 59-residue minimal region that is conserved in all BRC motifs. These data suggest that the BRC repeats function to bind RAD51.

Germline mutations in the brca1 and brca2 tumor suppressor genes account for approximately 5-10% of all breast cancer cases (1)(2)(3)(4). In addition, deleterious alleles of brca1 or brca2 are responsible for almost all familial ovarian cancer, and deleterious alleles of brca2 are involved in hereditary male breast cancer (1)(2)(3)(4). Currently the mechanism of action of these two genes remains largely undefined. The brca1 and brca2 genes encode large proteins, 1863 and 3418 amino acids, respectively (1,3). A search of the public sequence data bases has revealed little sequence homology to previously identified proteins. However, analysis of the protein sequence has revealed several statistically significant repeated motifs in these genes (5,6). For example, eight internal repeats, known as BRC motifs, are found clustered in exon 11 of the human BRCA2 protein. These motifs are not found in BRCA1 but are conserved in all mammalian BRCA2 proteins that have been sequenced. A similar motif is also present in the hypothetical protein encoded by the Caenorhabditis elegans gene T07E3.5 (5,7).
A number of studies have been conducted in the past year to elucidate the biological roles of the brca1 and brca2 genes.
Knock-outs of the mouse homologs of these genes indicate that they are essential during development. The ablation of either brca1 or brca2 results in an embryonic lethal phenotype characterized by failure of proliferation during approximately days 6 -8 of gestation (8,9). Levels of RNA expression from brca1 and brca2 are coordinately regulated during proliferation and differentiation in mammary epithelial cells (11,12). Furthermore, their expression varies at different stages of the cell cycle, with RNA levels peaking at the G 1 /S boundary (13,14). These findings provide preliminary evidence that these breast cancer genes participate in a common functional pathway. Three independent studies have now established that the RAD51 DNA repair protein is linked to the BRCA1 and BRCA2 pathways (9,10,15). Immunoprecipitation experiments reveal that RAD51 forms a complex with BRCA1, and immunofluorescence analysis shows that both proteins are co-localized to the synaptonemal complex of mouse meiotic chromosomes (15). The direct interaction of murine RAD51 with murine BRCA2 has been demonstrated in a yeast two-hybrid assay (9,10). The loss of interaction of these two proteins probably accounts for the hypersensitivity to ␥ irradiation that is observed in mouse embryos deficient in BRCA2 (9). rad51, a homolog of the Escherichia coli recA gene, is known to function in recombination and DNA repair (16,17). It is currently unclear how the interaction of RAD51 with either BRCA1 or BRCA2 affects its function. Alteration of normal DNA repair function may lead to genomic instabilities that eventually contribute to tumorigenesis.
We have conducted yeast two-hybrid searches using various segments of the human brca2 gene as "baits" and identified a number of interacting proteins, one of which is RAD51. To further understand the structural and functional relationships between human BRCA2 and RAD51, we map the minimal regions of the BRCA2 protein that mediate binding to RAD51. Specifically we show that RAD51 interacts with the eight BRC motifs of the human BRCA2 protein. These sites of interactions are distinct from those that have been previously reported for the murine BRCA2 and RAD51 proteins. GAAATCAACTAGTGAC; motif 4R, GGGTCGACTTCTTTACACTTTG-GGGCAGCTGTG; motif 5F, GGGAATTCGAAACAGCAAAAAGTCCT-GCAACTTG; motif 5R, GGGTCGACATAAGTATCTTGTTTTTCGGAG-AGATG; motif 6F, GGGAATTCCCCTGCAAAAATAAAAATGCAGCCA-TTA; motif 6R, GGGTCGACATGTGAATGCGTGCTACATTCATCAT-TA; motif 7F, GGGAATTCTTGGAAACTTCAGATATATGTAAATGT-AG; motif 7R, GGGTCGACTAAATGTTCTGGAGTACGTATAGCAG; motif 8F, GGGAATTCATACGTACTCCAGAACATTTAATATCC; motif 8R, GGGTCGACGTGCTCTGGGTTTCTCTTATCAACA; C. elegans motif F, GGGAATTCGTAAAAGACAGTTTCGACACAA; and C. elegans motif R, GGGTCGACTGGAGCTTCGAAATCTTCGAAACC.
The PCR products were subcloned into both pGBT.C and pGAD.C using standard techniques (23). Similarly, various truncated segments of RAD51 were created by PCR for subcloning. The sequence of all fusion constructs were verified by sequencing.
In Vitro Protein-Protein Interaction Assay-Templates for making radiolabeled proteins were either from PCR-generated DNA fragments or recombinant plasmids. The primers T7-act-forward (TAATACGAC-TCACTATAGGGAGACCACATGGATAAAGCGGAATTAATTCCCGA) and Adh-act-reverse (CCCTACATCATCATGCAGTATCTACGATTCA-TAGATCTCTG) specific for the pGAD.C vector were used to produce PCR templates of the BRC motifs and all other inserts that were fused to the Gal4p activation domain (24). The T7-act-forward primer contains the T7 RNA polymerase promoter. The PCR DNA templates were transcribed in vitro by T7 RNA polymerase using the Ribomax Kit (Promega) and translated in vitro in the presence of [ 35 S]methionine using rabbit reticulocyte lysate (Promega). In some experiments, the Single Tube Protein System 2 from Novagen was used to synthesize radiolabeled proteins.
GST fusion protein constructs were made by cloning the PCR products containing amino acid residues 1-44, 1-95, or 98 -339 of human RAD51 and residues 1912-2126 of BRCA2 into the GST expression vector pGEX (Pharmacia Biotech Inc.). All GST and GST fusion constructs were transformed into E. coli BL21(DE3)pLysS competent cells (Stratagene). The transformants were induced by isopropylthiogalactopyranoside, harvested, and lysed by sonication in phosphate-buffered saline with 0.1% Triton X-100. The binding assay of GST fusion protein with the radiolabeled protein was performed according to the protocol of Sande and Privalsky (25). Equal amounts of GST and GST fusion proteins, as determined by Coomassie Blue staining of aliquots run on gels, were used in each assay. The protein complexes bound to the beads were washed extensively and released by heating to 90°C in an equal volume of 2 ϫ SDS-PAGE loading buffer. The radiolabeled proteins were resolved by SDS-PAGE and visualized using the Molecular Dynamics Storm PhosphorImager System (Molecular Dynamics).

Interaction of the Internal Region of BRCA2 with RAD51-
The yeast two-hybrid system provides a powerful approach for identifying protein-protein interactions in vivo. In our initial experiments, we cloned the DNA segment of brca2 encoding amino acid residues 1912-2126 as a Gal4p DNA-binding domain fusion (Table I). This clone was used as a "bait" to search four different human Gal4p activation domain libraries. From a human B-cell library, we obtained 115 His ϩ , blue clones of approximately three million transformants. Sequencing of these potentially positive clones indicated that 22 of 115 were derived from the rad51 gene. The shortest of the RAD51 clones contained the cDNA segment encompassing amino acids 98 -339. This result indicates that the N-terminal portion of RAD51 is not necessary for binding amino acid residues 1912-2126 of the human BRCA2 protein. However, while our work was in progress, it was reported that the N terminus of the murine RAD51 protein is required for binding the C-terminal portion of the murine BRCA2 protein (9). For this reason, we also examined several overlapping segments that encompassed most of the BRCA2 gene to determine if additional regions of BRCA2 could interact with RAD51. Several BRCA2 segments as Gal4p DNA-binding domain fusions and RAD51 as a Gal4p activation domain fusion were co-expressed in yeast cells. The results of ␤-galactosidase filter assays are summarized in Table I. Four regions (amino acid residues 736 -1253, 1253-1708, 1708 -2099, and 1912-2126) of the BRCA2 protein scored pos-itive for ␤-galactosidase activity, indicating an interaction with RAD51. Two regions of BRCA2, encompassing residues 1253-1708 and 1708 -2099, when expressed as DNA-binding domain fusions, gave rise to light blue color. However, the intensity of the blue color increased dramatically when the RAD51 activation domain fusion was present, indicating an interaction with RAD51.
Interaction of the BRC Motifs with RAD51-Eight BRC motifs are found in exon 11 (encoding amino acid residues 638 -2280) of the human brca2 gene (5,7). Interactions of RAD51 with fragments of BRCA2 encompassing residues 736 -2126 raised the possibility that the BRC motifs might serve as specific sites for RAD51 interaction. To test this hypothesis, each BRC motif was tested individually for interaction with RAD51 in a yeast two-hybrid assay. As shown in Table II, six of the eight motifs (BRC1, -2, -3, -4, -7, and -8) interacted with RAD51 when it was expressed as an activation domain fusion. Because the context of folding in the Gal4p-DNA-binding domain fusion might have affected the binding of motifs 5 and 6 to RAD51, we performed swapping experiments in which motifs 5 and 6 were engineered as activation domain fusions and RAD51 was engineered as a DNA-binding domain fusion. After co-transformation, motifs 5 and 6 remained negative in the ␤-galactosidase assay (Table II). We also examined the BRC motif from C. elegans to see if it was able to interact with RAD51 in a similar yeast two-hybrid assay. A positive interaction with RAD51 was observed (Table II).
In Vitro Protein-Protein Interaction between BRCA2 and RAD51-In vitro biochemical assays provide an alternative means to investigate protein-protein interaction. Different segments of the RAD51 protein, residues 1-44, 1-95, and 98 -339, cloned as glutathione S-transferase fusions, were examined for in vitro association with radiolabeled BRCA2. The BRCA2 segment encompassing residues 1912-2126 was first used to optimize our assay. It was found that this region interacts with GST-RAD98 -339 but not with nonrecombinant GST, GST-RAD1-44, or GST-RAD1-95 (Fig. 1A). In a negative control, radiolabeled ␤-galactosidase was not retained on beads coupled to GST or GST-RAD51 fusions (data not shown). This assay was also used to confirm the physical associations of RAD51 with the BRC motifs of human BRCA2 and C. elegans. It was found that only amino acid residues 98 -339 of RAD51 could associate with each of the radiolabeled BRC motifs (represent-  Fig. 1, B, C, and D). Interestingly, BRC motifs 5 and 6, which scored negative in the yeast twohybrid assay, appeared to interact significantly and specifically with RAD51 in the in vitro assay.

Minimal Region Required for RAD51 and BRC Motif
Interaction-To determine the minimal region of RAD51 that can interact with the BRC motifs, we cloned a series of truncated segments of RAD51 into pGBT.C and co-transformed them into yeast with the BRC motif 3 in pGAD.C. As summarized in Table III, the segment from residues 98 -339 of RAD51 is required for the interaction with BRC3, which indicates that most of the C-terminal part of RAD51 is crucial to binding. In addition, we performed the reciprocal experiment to determine the minimal region of a BRC motif necessary for binding to RAD51. The motif BRC3 was arbitrarily chosen for the deletion analysis. It was found that the amino acid residues from positions 1414 to 1463 of BRCA2 are critical to binding (Table III). This minimal region includes all the conserved amino acids that form the basis of the BRC consensus sequence (5,7). DISCUSSION The identification of components in the BRCA2 functional pathway should further our understanding of BRCA2 function and its role in tumorigenesis. Through yeast two-hybrid screens of human cDNA libraries, we identified RAD51 as a BRCA2-binding protein. Sharan et al. (9) previously demonstrated that the C-terminal region of murine BRCA2 interacts with the N-terminal portion of murine RAD51. We identified eight sites of interaction between these proteins. Specifically the eight BRC motifs encoded in exon 11 of the human brca2 gene interact with RAD51. In addition, a similar motif found in a C. elegans hypothetical protein also interacts with RAD51. These data suggest that the BRC motifs function to bind RAD51.
Sequence comparison of the BRCA2 proteins from six mammalian species have revealed sequences conserved among the eight BRC motifs, which strongly implies functional significance (5, 7). Our findings that RAD51 binds to different protein sequences encoded by the 5-kilobase exon 11 of BRCA2 led us to examine whether the BRC internal repeats are the actual sites of interactions with RAD51. Because the amino acid sequences of these motifs are not identical, we analyzed each BRC motif individually to determine if they could associate with RAD51. All eight human BRC motifs showed interaction in biochemical assays, but only six of eight BRC motifs bound specifically to RAD51 in the yeast two-hybrid assays. This discrepancy in binding specificity might be caused by the different protein context of Gal4p and GST fusions. Subtle changes in the amino acid sequence of BRC motif 5 and 6 could have affected the proper folding of the predicted globular domain in these motifs in yeast.
Deletion analyses of the BRC motif 3 and RAD51 identified the minimal regions in both proteins that are required for binding. Our results indicate that the minimal region defined for the BRC motif 3 encompasses the entire consensus sequence that is highly conserved among mammalian species (7). Similarly, the minimal region of human RAD51, amino acid   residues 98 -339, that interacts with human BRCA2 is conserved among prokaryotic and eukaryotic species. The homologous region in the E. coli recA protein has been demonstrated to contain ATPase activity and is involved in oligomer formation and recombination (26). The role of multiple RAD51-binding sites in BRCA2 is open to interpretation. It seems reasonable that BRCA2 might bind several molecules of RAD51 simultaneously and therefore might serve as a scaffold for RAD51 assembly during DNA repair and recombination. Alternatively the multiple binding sites might play a role in the sequestration of active RAD51 or the facilitation of functional interactions of RAD51 with itself or other proteins. Because a similar BRC motif in a hypothetical protein from C. elegans is capable of binding to human RAD51, this same structural and functional relationship may hold in other eukaryotic species.
Using biochemical assays, we have confirmed that the association between RAD51 and BRCA2 is direct (10). In contrast, the binding of RAD51 to BRCA1 may be mediated through another protein in a complex (15). We have searched extensively for possible interactions between RAD51 and BRCA1 using the yeast two-hybrid assay, but so far without success (data not shown). It is interesting to note that the tumor suppressor gene p53 also associates with RAD51 (27). Together these data strongly suggest that RAD51 represents a crucial functional junction linking recombination and repair to cell cycle checkpoint signaling.
Our protein-protein interaction data, coupled with the previous observations that 1) brca2-null mouse embryos are hypersensitive to ␥ irradiation, 2) brca1 and brca2-null mouse embryos drastically overexpress p21 and suffer a failure of embryonic cell proliferation, and 3) there is a direct proteinprotein interaction between RAD51 and p53, raise a critical question about the function of BRCA1 and BRCA2 (28 -31). Is the primary deficit in brca1-null and/or brca2-null tumors a failure of DNA repair, or is it a failure of signal transduction associated with the need for DNA repair? The answer to this question could have bearings on the rationale for use of brca1 and brca2 as anti-tumor gene therapy agents in patients. The restoration of BRCA1 and BRCA2 function might stabilize genome integrity in the tumor without reversing the damages that have already occurred. In contrast, restoration of signal transduction might render the tumors more sensitive to apoptosis, a process that can be triggered by p53 or possibly chemotherapeutic agents. Further research is needed to address these possibilities.