Overexpression of the DNA double-strand break (DSB) repair gene RAD51 was identified as part of a 13-gene signature predictive of rapid brain metastasis development (≤3 years vs. >3 years after diagnosis, HR=3.03, p=0.0008) in primary tumors of chemotherapy-naïve HER2+ metastatic breast cancer patients. We also found RAD51 mRNA overexpressed (1.5-fold, p=0.001) in resected brain metastases compared to primary breast tumors from the same patients. In addition, RAD51 expression was significantly higher (1.4-fold, p=0.008,) in brain metastases compared to systemic (bone/lung) metastases. We therefore hypothesize that RAD51 promotes brain metastasis development in breast cancer patients. Given the known role of RAD51 in homologous recombination, important for the repair of DNA DSBs, it was unknown whether RAD51 would promote brain metastasis of breast cancer only in the presence of DNA-damaging chemotherapy or, alternatively, have endogenous biological functions in tumor cells. To functionally evaluate whether RAD51 promotes brain metastases, RAD51 was overexpressed at physiologically relevant levels in a brain-metastatic variants of triple-negative MDA-MB-231 breast carcinoma (231-BR) cells. In vitro, this resulted in a ∼30% increase in clonogenic colony formation (p=0.004), without affecting cell proliferation rates. RAD51 overexpression also conferred chemoresistance to DNA DSB inducing agents such as 25uM carboplatin (p=0.03) and 25nM doxorubicin (p=0.02), but not to agents that do not cause DNA DSBs, such as 2.5nM paclitaxel. In vivo, overexpression of RAD51 in 231-BR cells increased formation of large brain metastases, the equivalent of clinically detectable lesions, (4.0 and 3.3-fold, p=0.01 and p=0.005) and micro-metastases (3.1 and 3.5-fold, p=0.01 and p=0.001) in two independent xenograft experiments. Importantly, this brain-metastasis promoting effect occurred without drugs to induce DSB. This effect was recapitulated in a second animal model based on a brain-metastatic subline of 4T1 murine mammary carcinoma cells (4T1-BR). In this model, RAD51 overexpression increased large brain metastases (3.3-fold, p=0.02) and micro-metastases (2.0-fold, p=0.04) compared to vector control. ShRNA-mediated RAD51 knockdown in 4T1-BR cells resulted in the opposite effect, reducing large brain metastases (1.8-fold, p=0.1) and micro-metastases (2.5-fold, p=0.04). Together these data suggest a driving role for RAD51 in the development of brain metastases from breast cancer.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 5306. doi:1538-7445.AM2012-5306