Electrophilic fatty acids impair RAD51 function and potentiate the effects of DNA-damaging agents on growth of triple-negative breast cells

Homologous recombination (HR)-directed DNA double-strand break (DSB) repair enables template-directed DNA repair to maintain genomic stability. RAD51 recombinase (RAD51) is a critical component of HR and facilitates DNA strand exchange in DSB repair. We report here that treating triple-negative breast cancer (TNBC) cells with the fatty acid nitroalkene 10-nitro-octadec-9-enoic acid (OA-NO2) in combination with the antineoplastic DNA-damaging agents doxorubicin, cisplatin, olaparib, and γ-irradiation (IR) enhances the antiproliferative effects of these agents. OA-NO2 inhibited IR-induced RAD51 foci formation and enhanced H2A histone family member X (H2AX) phosphorylation in TNBC cells. Analyses of fluorescent DSB reporter activity with both static-flow cytometry and kinetic live-cell studies enabling temporal resolution of recombination revealed that OA-NO2 inhibits HR and not nonhomologous end joining (NHEJ). OA-NO2 alkylated Cys-319 in RAD51, and this alkylation depended on the Michael acceptor properties of OA-NO2 because nonnitrated and saturated nonelectrophilic analogs of OA-NO2, octadecanoic acid and 10-nitro-octadecanoic acid, did not react with Cys-319. Of note, OA-NO2 alkylation of RAD51 inhibited its binding to ssDNA. RAD51 Cys-319 resides within the SH3-binding site of ABL proto-oncogene 1, nonreceptor tyrosine kinase (ABL1), so we investigated the effect of OA-NO2–mediated Cys-319 alkylation on ABL1 binding and found that OA-NO2 inhibits RAD51–ABL1 complex formation both in vitro and in cell-based immunoprecipitation assays. The inhibition of the RAD51–ABL1 complex also suppressed downstream RAD51 Tyr-315 phosphorylation. In conclusion, RAD51 Cys-319 is a functionally significant site for adduction of soft electrophiles such as OA-NO2 and suggests further investigation of lipid electrophile–based combinational therapies for TNBC.

The multitude of exogenously and endogenously stimulated DNA-damaging events requires that DNA damage be vigilantly detected and efficiently repaired. Several DNA repair mechanisms have been identified that ameliorate deleterious genomic perturbations such as direct reversal, mismatch repair, nucleotide excision repair, base excision repair, and double-stranded break (DSB) repair (4). DNA DSBs are particularly pathogenic as the loss of genomic material and mutations promote genomic variability and disequilibrium. There are two main pathways to repair DSBs: nonhomologous end joining (NHEJ) and HR. Although NHEJ is faster and more frequently used, HR repair mechanisms maintain the highest fidelity of the genome. HR repair protects cells from the deleterious genomic instability caused by DSB by correcting for genetic material loss through homologous template searches that maintain the genomic landscape (5). RAD51 is a critical component of HR that facilitates the homology search and strand exchange to repair DSBs (6). RAD51 and the structurally similar proteins XRCC2, XRCC3, RAD51B, RAD51C, RAD51D, DMC1, and SWSAP1 all work in concert to promote HR (7). Consequently, reductions in RAD51 and paralog activity are linked with carcinogenesis (8).
Although RAD51 is essential for high-fidelity repair of DSB to maintain genomic homeostasis, overexpression of RAD51 in cancer can also have detrimental consequences. The extent of RAD51 overexpression is correlated with breast cancer tumor grade, and RAD51 overexpression has been identified in TNBC cell lines and metastatic patient samples (9,10). Overexpression of RAD51 inhibits chemotherapeutic efficacy in cancer patients by rendering cancer cells more resistant to DNA-damaging agents. Responses to neoadjuvant chemotherapy are inversely correlated with BRCA1, ␥H2AX, and RAD51 foci before treatment, as well as the numbers of RAD51 foci following treatment (11,12).
We report that combination treatments of OA-NO 2 with the antineoplastic agents doxorubicin, cisplatin, IR, and olaparib enhance the antiproliferative effect of these DNA-damaging therapeutic strategies in TNBC cells. OA-NO 2 was identified to suppress IR-induced RAD51 foci formation, inhibit RAD51 binding to ssDNA, decrease HR, induce phosphorylation of Ser-139 H2AX (␥H2AX), disrupt RAD51-ABL heterodimerization, and decrease RAD51 Tyr-315 phosphorylation. These observations reinforce the concept that reactive species induce genomic perturbations in part via the disruption of HR and reveal a novel therapeutic strategy: that redox-derived soft electrophiles sensitize cancer cells to DNA-directed therapeutic strategies such as IR, cisplatin, and doxorubicin.

OA-NO 2 inhibits TNBC cell growth, RAD51 foci formation, and sensitivity to ionizing radiation
Current data indicate that OA-NO 2 inhibits multiple aspects of TNBC epithelial cell, but not nontumorigenic breast epithelial cell, NF-B signaling, by alkylating functionally significant thiols in (a) the inhibitor of NF-B subunit kinase ␤, thus lim-iting downstream IK␣ phosphorylation and (b) the NF-B RelA protein, thus preventing DNA binding and promoting RelA polyubiquitination and proteasomal degradation (3). This motivated assessing whether OA-NO 2 could also enhance TNBC DNA damage in vivo. MDA-MD-231 cells were implanted into the mammary gland of mice, and when tumors reached a volume of 100 mm 3 , mice were treated with 15 mg/kg nonelectrophilic fatty acid oleic acid (OA) or OA-NO 2 by gavage for 4 weeks. Mice treated with OA-NO 2 had significantly decreased tumor growth rates when compared with OAtreated controls ( Fig. 1A and Fig. S1A). Probing tumor levels of the DNA damage biomarker ␥H2AX by immunoblotting showed that OA-NO 2 -treated mice displayed higher levels of ␥H2AX (Fig. 1B). Densitometric quantification of tumoral ␥H2AX/␤-actin protein levels in OA-and OA-NO 2 -treated mice showed increased ␥H2AX in OA-NO 2 -treated mice. This response became statistically significant after Grubbs outlier detection and elimination of OA-treated mouse 2 (Fig. S1, B and C). The orthotopic tumor in OA mouse 2 was the largest tumor in the study, with necrosis potentially causing enhanced ␥H2AX levels.
The TNBC growth-inhibitory effects of OA-NO 2 were then evaluated in combination with DNA-damaging agents. The cell lines MDA-MB-231, BT-549, and Hs578T were treated with OA-NO 2 daily for 3 days, and relative cell numbers were quantified by measuring the ATP-dependent luminescence signal generated using Ultra-Glo luciferase with the substrate luciferin. The EC 50 values for growth inhibition of TNBC cells ranged from 1.98 Ϯ 0.52 (Hs578T) to 3.78 Ϯ 0.48 M (BT-549) with MDA-MB-231 cells displaying an EC 50 value of 3.66 Ϯ 0.14 M (Fig. 1C and Fig. S1D). We next tested the DNA-damaging agents doxorubicin and cisplatin in combination with daily treatment with 2 M OA-NO 2 . OA-NO 2 enhanced growth inhibition of doxorubicin in MDA-MB-231 and Hs578T cells, by 7-and 5-fold, respectively ( Fig. 1D and Fig. S1E). The growth of BT-549 cells was not affected. Cotreatment with OA-NO 2 and cisplatin showed a similar trend for MDA-MB-231 and Hs578T cells, which displayed increased growth inhibition by 6-and 3-fold, respectively, whereas growth inhibition of BT-549 cells was suppressed 1.4-fold ( Fig. 1E and Fig. S1F). A subset of TNBC cells are sensitive to PARP inhibition and display a BRCAness phenotype in the presence of WT BRCA1 (14), so the PARP-1 inhibitor olaparib was evaluated to determine whether a combination treatment with OA-NO 2 enhanced potency. MDA-MB-231, Hs578T, and BT-549 cells all displayed enhanced growth inhibition when olaparib was combined with OA-NO 2 by 5-, 17-, and 3-fold, respectively ( Fig. 1F and Fig. S1G). To specifically show that olaparib treatment in combination with OA-NO 2 , but not OA, affected proliferation and to evaluate the effect of daily olaparib media exchanges, MDA-MB-231 cell proliferation was quantified in dose-response assays. Daily administration of olaparib in combination with OA did not significantly alter the EC 50 of olaparib in MDA-MB-231 cells. In contrast, olaparib in combination with OA-NO 2 significantly inhibited growth (Fig. 1G). Thus, standard TNBC chemotherapeutic drugs as well as targeted PARP-1 inhibition exhibited enhanced antiproliferative effects when coadministered with OA-NO 2 in TNBC cells.

ACCELERATED COMMUNICATION: OA-NO 2 targets HR in breast cancer
The heightened tumor ␥H2AX levels in vivo and sensitization of TNBC cells to DNA-damaging agents, especially in the context of olaparib-induced responses, led to further exploration of DNA damage repair modulation by OA-NO 2 . As olaparib sensitivity is a hallmark of HR-deficient cells (15), whether OA-NO 2 impacted DNA repair by HR was evaluated. To specifically probe DNA double-strand break repair, MDA-MB-231 cells were challenged with 5 Gy of IR, and RAD51 foci were quantified. The treatment of breast cancer cells with OA-NO 2 inhibited RAD51 foci formation as reflected by (a) the number of cells with more than five foci and (b) responses of vehicletreated cells following IR (Fig. 1H). Cell cycle analysis of MDA-MB-231 cells confirmed that no significant changes to the cell cycle occurred that might indirectly alter RAD51 foci formation (Fig. S1H). Evaluation of nuclear ␥H2AX staining to probe for DNA damage of MDA-MB-231 cells in the presence or absence of 5 Gy of IR found that OA-NO 2 significantly increased nuclear ␥H2AX localization in irradiated MDA-MB-231 cells compared with vehicle controls, indicating increases in DSBs and overall DNA damage (Fig. 1I). Increasing concentrations of OA-NO 2 also enhanced breast cancer cell death in a clonogenic assay following irradiation with 2 Gy of IR (Fig. S1I). Evaluation of the DNA-damaging effects of OA and OA-NO 2 on nontransformed MCF10A cells following 5 Gy of IR found that nuclear ␥H2AX staining was only increased in MDA-MB-231 TNBC cells treated with OA-NO 2 and not MCF10A cells (Fig. S2, A and B).

OA-NO 2 decreases HR but not NHEJ efficiency
OA-NO 2 -dependent effects on HR DNA repair were further investigated by utilizing a direct repeat GFP (DR-GFP) reporter assay. This analysis quantifies intracellular recombination of an integrated cDNA cassette of two tandem nonfluorescent GFP constructs, following introduction of an I-SceI cleavage to the system, by measuring the fluorescent GFP protein that is produced following successful recombination (16). Daily OA-NO 2 treatment of U2OS cells harboring the DR-GFP construct revealed that after I-SceI transfection, the number of GFP-positive cells was significantly decreased by 2-fold when compared with native OA or vehicle control after 48 h ( Fig. 2A and Fig.  S3A). A novel strategy was used to measure the kinetics of changes in HR in live cells by using automated fluorescence  (Fig. 2B). The impact of OA-NO 2 on suppression of DSB repair through both the HR and NHEJ pathways was examined by utilizing an EJ5-GFP NHEJ reporter assay, which separates GFP cDNA from a transcriptional promoter with a puromycin resistance gene flanked by two I-SceI cleavage sites (17). In contrast to the effects seen by DR-GFP-mediated HR measurements, EJ5-GFP U2OS cells showed no effect of OA-NO 2 on NHEJ. This was indicated by an absence of changes in the number of GFP-positive cells following I-SceI cleavage after 48 h by flow cytometric analysis or over 68 h by live-cell fluorescence microscopy (Fig. 2, C and D, and Fig. S3C).

OA-NO 2 targets RAD51 Cys-319 and decreases RAD51 phosphorylation
Inhibition of IR-induced RAD51 foci formation and DR-GFP HR reporter functionality by OA-NO 2 was further studied by testing whether overexpression of the critical HR protein RAD51 in the HR reporter cells could rescue the effects of OA-NO 2 . HR activity, as measured by the percentage of GFPpositive cells relative to OA control treatment, was significantly increased in U2OS DR-GFP reporter cells stably overexpressing RAD51 treated with 5 M OA-NO 2 compared with control reporter cells ( Fig. 3A and Fig. S4A). Protein structural data (Protein Data Bank (PDB) code 1N0W) show that Cys-319 is a solvent-exposed nucleophile within the RAD51 C terminus (Fig. S4B) that is susceptible to reaction with RI-1, a reagent also having Michael acceptor qualities (18). Moreover, fluorophore adduction of Cys-319 disrupts RAD51 filament formation in vitro (19). It was hypothesized that OA-NO 2 would react with RAD51 Cys-319. Indeed, biotin-OA-NO 2 , but not the nonelectrophilic biotin-OA and biotin-10-nitro-octadecanoic acid (SA-NO 2 ), supported affinity precipitation of RAD51 from cell lysates with streptavidin-labeled beads (Fig. 3B). Comparing RAD51 C312S or C319S mutant reaction with biotin-OA-NO 2 revealed a preferential reaction of OA-NO 2 with Cys-319 (Fig. 3C). RAD51 C312S and RAD51 WT controls were readily affinity-precipitated by biotin-OA-NO 2 as opposed to when RAD51 C319S was expressed in mutant cells. Of note, the RAD51 C312S mutant displayed enhanced precipitation of OA-NO 2 , which may reflect interruption of a disulfide bond between RAD51 Cys-312 and Cys-319 or another intracellular protein that obscures Cys-319. The ability of OA-NO 2 to specifically disrupt RAD51 binding from DNA was probed by quantifying changes in fluorescence polarization of an Alexa Fluor 488 -conjugated single-stranded oligonucleotide in vitro. OA-NO 2 , but not OA, decreased the relative polarization of RAD51 in the presence of ATP and DNA (Fig. 3D). Control experiments found that OA and OA-NO 2 did not cause nonspecific effects through fluorophore quenching to decrease fluorescence polarization (Fig.  S4C). Computational analysis revealed that OA-NO 2 alkylation of RAD51 Cys-319 is further stabilized by hydrophobic interactions with Pro-318 of RAD51 and hydrogen bonding with Glu-322 (Fig. 3E).

ACCELERATED COMMUNICATION: OA-NO 2 targets HR in breast cancer
Cys-319 is located in the RAD51 C terminus within one of the two ABL-SH3-binding domains (amino acids 283-286 and 318 -321) (20). In addition to RAD51 filament disruption, OA-NO 2 inhibited heterodimerization of RAD51 and ABL. IP analysis revealed that purified RAD51 and catalytic ABL core (Src homology 2 (SH2), SH3, and kinase domains only) complex formation was abolished by OA-NO 2 (Fig. 3F). ABL regulates RAD51 activity via sequential phosphorylation of RAD51 Tyr-54 and then Tyr-315 (21,22). By transfecting FLAG-RAD51 and ABL core into 293T cells, the impact of OA-NO 2 on RAD51-ABL complex formation and RAD51 Tyr-315 phosphorylation was examined. After treating cells with 0 -5 M OA-NO 2 for 1 h, FLAG IP analysis revealed that OA-NO 2 decreased the amount of ABL bound to RAD51 (Fig. 3G). Along with the inhibition of RAD51-ABL complex formation, RAD51 Tyr-315 phosphorylation was also inhibited by OA-NO 2 in FLAG-RAD51-and ABL-expressing cells. To define whether OA-NO 2 alkylates endogenous RAD51 in TNBC cells, biotin-OA-NO 2 was added to MDA-MB-231 and MDA-MB-468 cells. Biotin-OA-NO 2 -RAD51 complex formation upon streptavidin precipitation was observed in lysates of both cell lines (Fig.  3H). Overall, these data reveal that OA-NO 2 inhibited HR by forming adducts with RAD51 and possibly additional HR-related target proteins to enhance sensitivity to DNA-directed cancer therapies (Fig. 3I).

Discussion
Fatty acid nitroalkenes are endogenously produced products of nitric oxide-and nitrite-dependent nitration of unsaturated fatty acids. By virtue of kinetically rapid and reversible Michael addition, fatty acid nitroalkenes mediate the post-translational modification of susceptible Cys residues of proteins, in some cases modifying protein function and inducing signaling responses via pleiotropic mechanisms (1, 2, 23, 24). The present , and fluorescence polarization was quantified and normalized to a control lacking ATP. E, molecular modeling of RAD51 (blue) and OA-NO 2 (purple). Binding of OA-NO 2 with the Cys-319 residue (gold) of RAD51 is predicted to be further stabilized by hydrophobic interactions with Pro-318 and possible hydrogen bonding with Glu-322 of RAD51. F, OA-NO 2 disrupts ABL binding to RAD51 in vitro. Purified RAD51 and ABL core proteins were incubated with OA-NO 2 at 0, 100, or 500 nM for 1 h, and ABL was precipitated. The amount of bound RAD51 was detected by immunoblotting. G, OA-NO 2 disrupts RAD51 and ABL interactions by IP and reduces RAD51 Tyr-315 phosphorylation. 293T cells were transfected with FLAG-RAD51 and ABL core protein and then treated with OA-NO 2 for 1 h. RAD51 interactions with ABL and phosphorylated RAD51 Tyr-315 were probed by IP and immunoblotting. H, OA-NO 2 binds RAD51 in MDA-MB-231 or MDA-MB-468 cells. Cells were incubated with biotinylated OA-NO 2 , and then lysates were precipitated with streptavidin-coated agarose and detected by immunoblotting. I, OA-NO 2 decreases HR and causes genomic instability and death in TNBC cells. All values indicate average, and error bars represent S.E.; n ϭ 3.

ACCELERATED COMMUNICATION: OA-NO 2 targets HR in breast cancer
results indicate that OA-NO 2 decreases the proliferation of TNBC cells, especially when coadministered with clinically relevant DNA-directed therapeutic agents (Fig. 1). OA-NO 2 also amplified the induction of DSB through IR or I-SceI DNA cleavage by limiting IR-induced nuclear RAD51 foci formation and DNA recombination, specifically via inhibiting HR and not NHEJ (Figs. 1 and 2). The functionally significant Cys-319 of RAD51 was targeted by OA-NO 2 , but not by nonelectrophilic native and nitroalkane-substituted control fatty acids (Fig. 3). Cys-319 alkylation by OA-NO 2 disrupted RAD51 dimerization with ABL and decreased ABL-induced phosphorylation of RAD51 Tyr-315 (Fig. 3). OA-NO 2 may also target proteins beyond RAD51 that modulate HR.
The potent antiproliferative effect of olaparib, when administered in combination with OA-NO 2 in TNBC cell lines, indicates the pharmacological induction of a BRCAness phenotype by OA-NO 2 (Fig. 1). Suppression of HR-mediated DNA DSB repair by OA-NO 2 is reflective of loss of function mutations in BRCA genes, which cause deficits in DNA repair capacity via impairment of HR. Breast cancer patients harboring BRCA1 loss of function mutations may also benefit from suppression of RAD51, as increased expression of RAD51 bypasses BRCA1 function and is a common feature of BRCA1-deficient breast tumors (25). Additional investigation of further OA-NO 2 targets and actions, coupled with PARP inhibition in BRCA1-deficient breast cancer backgrounds, is thus warranted.
Although functional HR is important for maintaining genome stability, an enhancement of homology-directed DNA repair activities impedes chemotherapeutic and ionizing radiation treatments for cancer (6,7). The elevated expression of RAD51 is positively correlated with breast cancer tumor grade and has been identified in several TNBC cell lines and metastatic patient samples (9,10). Several studies have attempted to harness RAD51 inhibition to promote lethality in cancer cells. Inhibition of RAD51 with small-molecule inhibitors can sensitize cancer cells to chemotherapeutic agents or IR (e.g. DIDS (26), B02 (27,28), RI-1 (18), and IBR2 (29)). For example, RI-1 was identified in a high-throughput screen to potentiate RAD51 filament formation and HR activity, fortuitously adducting Cys-319 (18). Unfortunately, RI-1 has multiple electrophilic centers in a complex biphenolic morpholino structure, thus inducing irreversible Michael addition and an incompatibility for in vivo applications due to toxicity.
The RAD51 Cys-319 represents an important functional site within the RAD51 protein. Homomultimeric RAD51 filaments interface surrounding the Cys-319 residue, which is located within an SH3 domain and nearby an ATPase domain (PDB code 1N0W) (30). Post-translational thiol modification or pharmacological targeting of Cys-319 may thus disrupt RAD51 function through multiple mechanisms.
It was previously reported that OA-NO 2 more selectively targets TNBC thiols of the NF-B signaling pathway, as opposed to nontumorigenic breast epithelium, due to the more effective mechanisms for maintaining redox homeostasis in normal breast epithelium (3). We now identify another specific target of electrophilic nitroalkenes, the Cys-319 of RAD51, that upon alkylation inhibits RAD51 binding to ssDNA (Fig. 3). Thus, the administration of synthetic homologs of endogenously occur-ring fatty acid nitroalkenes offers a viable option for inactivating RAD51. The clinical administration of intravenous and oral formulations of OA-NO 2 (i.v. IND, 122583; oral IND, 124524) is safe, having cleared multiple Phase I and drug-drug interaction studies. An oral formulation of OA-NO 2 is now in multicenter Phase II trials for treating chronic inflammatory-related diseases. The present results motivate further investigation into whether endogenously generated soft electrophiles or their exogenously administered synthetic homologs might play a role in modulating DNA repair and other signaling responses that improve treatment of drug-resistant cancers.

OA-NO 2 in vivo
Animals used for this study were approved by and conducted according to the guidelines of the University of Pittsburgh Institutional Animal Care and Use Committee (protocol 18093443). MDA-MB-231 cells (0.5 ϫ 10 6 ) were injected into the mammary fat pad (left fourth gland) of 6-week-old female nude mice in a volume of 20 l of sterile saline. When tumors reached an average volume of 100 mm 3 , mice were randomized into groups and administered vehicle (tricaprylin) ϩ 15 mg/kg OA or vehicle ϩ 15 mg/kg OA-NO 2 every day by gavage (200 l) for 4 weeks. The surgical procedure has been described previously (32).

DSB repair assays
Measurements of HR and NHEJ assays were performed as described previously (16,34). HR activity was measured by counting GFP-positive cells by flow cytometry at the Magee-Womens Research Institute Flow Cytometry Core using a BD LSR II flow cytometer (BD Biosciences). RAD51-overexpressing cells were generated by stable transfection of pLVX-RAD51-IRES-Neo and selection with geneticin (Invitrogen).

Kinetic DSB repair assays
U2OS cells were prepared as above, but 5 h following compound treatment, cells were transferred into the Incucyte Zoom (Essen) live-cell imaging automated fluorescence microscope at 37°C with 5% CO 2 . Cell confluence and green object count per mm 2 were determined using Incucyte Zoom software. Green object count per field was normalized to cell confluence to correct for OA-NO 2 -induced effects on cell proliferation.

Immunostaining and imaging
To analyze RAD51 foci formation, 10,000 cells were plated to a CultureWell 16-well chambered coverglass (MIDSCI) coated with poly-L-lysine (Sigma) and incubated overnight in 5% FBS medium. Cells were then treated with OA-NO 2 and irradiated (Gammacell 40 Exactor ␥-Irradiator, Best Medical) with 5 Gy and incubated for 6 h. Cells were fixed with 10% formalin for 20 min at 4°C and immunostained with RAD51 (Santa Cruz Biotechnology) or ␥H2AX (EMD Biosciences) antibody. z-stack images were acquired using a Nikon A1R confocal microscope with 60ϫ oil objective, and acquisition was performed using NIS Elements software. Quantification of z-stacks and foci were completed using ImageJ software.

Cell cycle analysis
Cell cycle analysis was performed with propidium iodide on MDA-MB-231 cells treated with DMSO or 5 M OA-NO 2 (35). Samples were analyzed at the Magee-Womens Research Institute Flow Cytometry Core utilizing a BD LSR II flow cytometer.

Western blotting and immunoprecipitation
Cell lysates and immunoprecipitations were prepared as described previously (36). For immunoprecipitation analysis, one million HEK 293T cells were transiently transfected with FuGENE 6 (Promega) and 2 g of pQCXIP (empty vector) or FLAG-RAD51 pQCXIP plasmids and precipitated with anti-FLAG M2 affinity gel (Sigma).

Biotinylated OA-NO 2 affinity capture of RAD51
HEK 293T were transiently transfected with FuGENE 6 and 5 g of RAD51-expressing vectors (WT, C312S, or C319S). Cells were treated 24 h later with 5 M biotin-OA-NO 2 or biotin-SA-NO 2 for 1 h in 5% FBS medium. Cells were prepared as above. Precipitation of biotinylated OA-NO 2 protein adducts was accomplished with 8 l of streptavidin-agarose beads with 1 mg of total cell lysates incubated for 16 h at 4°C. Detection of RAD51 was accomplished by immunoblotting with RAD51 antibody (1:2,000) with actin antibody (1:3,000) probed as a loading control.

Protein purification for in vitro RAD51-ABL binding assay
Recombinant His-tagged RAD51 in the pET21a vector was transformed into Escherichia coli BL21(DE3)pLysS cells (EMD Millipore) and purified as described previously (37).

DNA binding assays
Reactions were performed in black 96-well plates (Greiner) in 50-l reaction volumes in 20 mM HEPES, pH 7.5, 10 mM MgCl 2 , 0.25 M BSA, 2% glycerol, 30 mM NaCl, and 4% DMSO. Purified RAD51 protein (Abcam) and OA (negative control) or OA-NO 2 were preincubated for 5 min at 25°C. 2 mM ATP and 100 nM 5Ј-Alexa Fluor 488 -conjugated ssDNA poly(dT) (Integrated DNA technologies) were added to the reaction and incubated for 90 min at 37°C. DNA binding was measured using fluorescence polarization on a Tecan Spark 20 M (excitation/ emission, 480 nm/535 nm). Compound fluorescence quenching was detected as above in the absence of RAD51 protein.

Molecular modeling
Structures for RAD51 (PDB code 1N0W (30) and OA-NO 2 were aligned using PyMOL 1.7.1. The structure of OA-NO 2 was generated using ChemDraw 15 (PerkinElmer Life Sciences) and converted to 3D structure using Open Babel version 2.3.1 (13).

Statistical analysis
Data represent the mean Ϯ S.E. from three independent experiments unless otherwise noted. A p value Ͻ0.05 was considered statistically significant. Nonlinear curves were generated in GraphPad Prism 7.0 (GraphPad Software, La Jolla, CA) for statistical analysis. EC 50 values and standard error were calculated from three independent experiments utilizing a nonlinear dose-response variable slope model. Significance was tested by one-way analysis of variance for multiple groups with Tukey post-test or by t test when groups were less than three. RAD51 foci number was analyzed with ImageJ. Nuclear boundaries were individually identified in more than 50 cells per treatment group in three independent experiments.