SUMMARYDNA double strand break (DSB) repair by Homologous recombination (HR) is initiated by the end resection, a process during which 3’ ssDNA overhangs are generated by the nucleolytic degradation. The extent of DNA end resection determines the choice of the DSB repair pathway. The role of several proteins including nucleases for end resection has been studied in detail. However, it is still unclear how the initial, nicked DNA generated by MRE11-RAD50-NBS1 is recognized and how subsequent proteins including EXO1 are recruited to DSB sites to facilitate extensive end resection. We found that the MutSβ (MSH2-MSH3) mismatch repair (MMR) complex is recruited to DSB sites by recognizing the initial nicked DNA at DSB sites through the interaction with the chromatin remodeling protein SMARCAD1. MSH2-MSH3 at DSB sites helps to recruit EXO1 for long-range resection and enhances its enzymatic activity. MSH2-MSH3 furthermore inhibits the access of DNA polymerase θ (POLQ), which promotes polymerase theta-mediated end-joining (TMEJ) of DSB. Collectively, our data show a direct role for MSH2-MSH3 in the initial stages of DSB repair by promoting end resection and influencing DSB repair pathway by favoring HR over TMEJ. Our findings extend the importance of MMR in DSB repair beyond established role in rejecting the invasion of sequences not perfectly homologous to template DNA during late-stage HR.

Background: Mutation carriers of BRCA1 and BRCA2 are well known to develop early onset breast cancer, with loss of the second allele occurring in the development of the tumor. However, by array comparative genomic hybridization (aCGH) studies, some sporadic breast cancers have a similar “genetic landscape” as BRCA-mutation carriers, showing large losses and gains across the genome. We have now identified that DNA repair defects involving the BRCA1-BRCA2 pathway can occur in the absence of mutations in either gene and in the absence of a deficiency in either protein. Methods: Fresh human breast cancer samples were irradiated, ex-vivo, to look for the ability to assemble RAD51 protein macrocomplexes or foci. Primary breast cancer specimens were obtained from consented patients with non-metastatic, invasive carcinomas following lumpectomy or mastectomy, without neoadjuvant cytotoxic or hormonal therapy. A single cell suspension was prepared from the tumor, with one half irradiated to 10Gy and the other half mock-treated. After 4h, cells were mounted, fixed on slides, and stained with anti-Rad51, anti-BRCA1, and anti-γH2AX antibodies. At least 200 nuclei were examined and scored using confocal microscopy. A failure to induce RAD51 nuclear foci by 2-fold after ionizing radiation was designated as defective in homologous recombination (HR). Results: For the 71 patient samples analyzed, we have 14 triple-negative tumors, of which 6 are HR-defective (42.8%); for Her2-amplified tumors, we have 6/19 (31.6%) that are HR-defective and for ER+ tumors 6/38 (15.8%). The overall incidence of HR-defective tumors is 18/71 (25.3%), which is substantially higher than we would have expected. Known mutation carriers were not included in the study, since these samples are BRCA-HR-defective in all cases we have tested. For the more recently acquired samples, we have undertaken additional tests to characterize the tumors: short-term growth assays in response to mitomycinC to validate that HR-defective tumors are indeed sensitive to cross-linking agents; and, a pilot analysis to study aCGH patterns in HR-defective tumors. The latter studies have compared 6 repair-deficient and 6 repair-proficient tumors using unsupervised cluster analysis of large block deletions or large block copy number increase, which clearly reveal that large block alterations are linked to repair-deficient tumors. Conclusions: There is a significant incidence of BRCA-HR defective sporadic breast cancers, as determined by RAD51 function in response to ionizing radiation plus genetic landscape alterations using aCGH. The pool of breast cancers that are susceptible to repair targeting strategies is larger than expected and is not readily defined by conventional diagnostic biomarker classification. These findings may account for the failure of the recent phase III study of the addition of iniparib to carboplatin and gemcitabine in triple-negative cancer as only a minority of the tumors will be susceptible to this targeting strategy. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr PD10-02.