Breaks in both DNA strands are a particularly dangerous threat to genome stability. At a DNA double-strand break (DSB), potentially lost sequence information cannot be recovered from the same DNA molecule. However, simple repair by joining two broken ends, though inherently error prone, is preferable to leaving ends broken and capable of causing genome rearrangements. To avoid DSB-induced genetic disinformation and disruption of vital processes, such as replication and transcription, cells possess robust mechanisms to repair DSBs. Because all breaks are not created equal, the particular repair mechanism used depends largely on what is possible and needed based on the structure of the broken DNA. We argue that although categorizing different DSB repair mechanisms along pathways and subpathways can be conceptually useful, in cells flexible and reversible interactions among DSB repair factors form a web from which a nonpredetermined path to repair for any number of different DNA breaks will emerge.

Defective homologous recombination (HR) DNA repair imposed by BRCA1 or BRCA2 deficiency sensitizes cells to poly (ADP-ribose) polymerase (PARP)-1 inhibition and is currently exploited in clinical treatment of HR-deficient tumors. Here we show that mild hyperthermia (41–42.5 °C) induces degradation of BRCA2 and inhibits HR. We demonstrate that hyperthermia can be used to sensitize innately HR-proficient tumor cells to PARP-1 inhibitors and that this effect can be enhanced by heat shock protein inhibition. Our results, obtained from cell lines and in vivo tumor models, enable the design of unique therapeutic strategies involving localized on-demand induction of HR deficiency, an approach that we term induced synthetic lethality.

EMBO Press is an editorially independent publishing platform for the development of EMBO scientific publications.

Homologous recombination is a versatile DNA damage repair pathway requiring Rad51 and Rad54. Here we show that a mammalian Rad54 paralog, Rad54B, displays physical and functional interactions with Rad51 and DNA that are similar to those of Rad54. While ablation of Rad54 in mouse embryonic stem (ES) cells leads to a mild reduction in homologous recombination efficiency, the absence of Rad54B has little effect. However, the absence of both Rad54 and Rad54B dramatically reduces homologous recombination efficiency. Furthermore, we show that Rad54B protects ES cells from ionizing radiation and the interstrand DNA cross-linking agent mitomycin C. Interestingly, at the ES cell level the paralogs do not display an additive or synergic interaction with respect to mitomycin C sensitivity, yet animals lacking both Rad54 and Rad54B are dramatically sensitized to mitomycin C compared to either single mutant. This suggests that the paralogs possibly function in a tissue-specific manner. Finally, we show that Rad54, but not Rad54B, is needed for a normal distribution of Rad51 on meiotic chromosomes. Thus, even though the paralogs have similar biochemical properties, genetic analysis in mice uncovered their nonoverlapping roles.

The high incidence of breast cancer has sparked the development of novel targeted and personalized therapies. Personalization of cancer treatment requires reliable prediction of chemotherapy responses in individual patients. Effective selection can prevent unnecessary treatment that would mainly result in the unwanted side effects of the therapy. This selection can be facilitated by characterization of individual tumors using robust and specific functional assays, which requires development of powerful ex vivo culture systems and procedures to analyze the response to treatment. We optimized culture methods for primary breast tumor samples that allowed propagation of tissue ex vivo. We combined several tissue culture strategies, including defined tissue slicing technology, growth medium optimization and use of a rotating platform to increase nutrient exchange. We could maintain tissue cultures for at least 7 days without losing tissue morphology, viability or cell proliferation. We also developed methods to determine the cytotoxic response of individual tumors to the chemotherapeutic treatment FAC (5-FU, Adriamycin [Doxorubicin] and Cyclophosphamide). Using this tool we designated tumors as sensitive or resistant and distinguished a clinically proven resistant tumor from other tumors. This method defines conditions that allow ex vivo testing of individual tumor responses to anti-cancer drugs and therefore might improve personalization of breast cancer treatment.

Off-target or random integration of exogenous DNA hampers precise genomic engineering and presents a safety risk in clinical gene therapy strategies. Genetic definition of random integration has been lacking for decades. Here, we show that the A-family DNA polymerase θ (Pol θ) promotes random integration, while canonical non-homologous DNA end joining plays a secondary role; cells double deficient for polymerase θ and canonical non-homologous DNA end joining are devoid of any integration events, demonstrating that these two mechanisms define random integration. In contrast, homologous recombination is not reduced in these cells and gene targeting is improved to 100% efficiency. Such complete reversal of integration outcome, from predominately random integration to exclusively gene targeting, provides a rational way forward to improve the efficacy and safety of DNA delivery and gene correction approaches.