RAD54B mutations enhance the sensitivity of ovarian cancer cells to poly(ADP-ribose) polymerase (PARP) inhibitors

Synthetic lethal targeting of homologous recombination (HR)–deficient ovarian cancers (OvCas) with poly(ADP-ribose) polymerase inhibitors (PARPis) has attracted considerable attention. Olaparib was the first PARPi approved by the Food and Drug Administration, offering significant clinical benefits in BRCA1/2-deficient OvCas. However, only approximately 20% of OvCa patients harbor BRCA1/2 mutations. Given the shared roles that BRCA1/2 have with other HR regulators, alterations in HR genes may also contribute to “BRCAness profiles” in OvCas. RAD54B has been considered a key player in HR repair, although its roles and therapeutic potential in cancers need further investigation. Here, we identified 22 frequently mutated HR genes by whole-exome sequencing of OvCa tissues from 82 patients. To our surprise, 7.3% of patients were found to harbor mutations of RAD54B, the third-highest mutated gene among patients. We determined that RAD54B-mutated tumor tissues harbored more DNA double-strand breaks than normal tissues. Additionally, we found that RAD54B knockdown inhibited HR repair, enhanced sensitivities of OvCa cells with increased DNA double-strand breaks to olaparib, and induced apoptosis. Enhanced inhibitory effects of olaparib on the growth of ES2 xenograft tumors were further demonstrated by RAD54B knockdown. Finally, we show that restoration with wildtype RAD54B rather than RAD54BN593S and RAD54BH219Y, identified in patients, abolished the effects of RAD54B knockdown, indicating these RAD54B mutants probably malfunctioned in HR repair. Our investigations may offer insight into the contributions of RAD54B mutations to synthetic lethality with olaparib treatment in OvCas, enrich the gene list for “HR deficiency scoring,” and expand the applications of PARPis.

Synthetic lethal targeting of homologous recombination (HR)-deficient ovarian cancers (OvCas) with poly(ADP-ribose) polymerase inhibitors (PARPis) has attracted considerable attention. Olaparib was the first PARPi approved by the Food and Drug Administration, offering significant clinical benefits in BRCA1/2-deficient OvCas. However, only approximately 20% of OvCa patients harbor BRCA1/2 mutations. Given the shared roles that BRCA1/2 have with other HR regulators, alterations in HR genes may also contribute to "BRCAness profiles" in OvCas. RAD54B has been considered a key player in HR repair, although its roles and therapeutic potential in cancers need further investigation. Here, we identified 22 frequently mutated HR genes by whole-exome sequencing of OvCa tissues from 82 patients. To our surprise, 7.3% of patients were found to harbor mutations of RAD54B, the third-highest mutated gene among patients. We determined that RAD54Bmutated tumor tissues harbored more DNA double-strand breaks than normal tissues. Additionally, we found that RAD54B knockdown inhibited HR repair, enhanced sensitivities of OvCa cells with increased DNA double-strand breaks to olaparib, and induced apoptosis. Enhanced inhibitory effects of olaparib on the growth of ES2 xenograft tumors were further demonstrated by RAD54B knockdown. Finally, we show that restoration with wildtype RAD54B rather than RAD54B N593S and RAD54B H219Y , identified in patients, abolished the effects of RAD54B knockdown, indicating these RAD54B mutants probably malfunctioned in HR repair. Our investigations may offer insight into the contributions of RAD54B mutations to synthetic lethality with olaparib treatment in OvCas, enrich the gene list for "HR deficiency scoring," and expand the applications of PARPis.
Synthetic lethality is one such anticancer therapeutic strategy, which takes advantages of lethal combination of two independently viable mutations (8). In OvCa therapy, this strategy exploits specific DNA repair-related genetic lesions in cells by targeting synthetic lethal partner genes. The synthetic lethal therapy can in principle result in selective OvCa cell death, while avoiding normal cells harmed. In HR-defective OvCas, poly(ADP-ribose) polymerases (PARPs) are considered as attractive synthetic lethal targets because of their crucial roles in DNA single-strand break repair (9,10). PARP inhibitors (PARPis) can inhibit PARP-dependent single-strand break repair, leading to accumulation of fatal double-strand breaks (DSBs) that causes synthetic lethality in HR-defective OvCa cells. The first PARPi approved by Food and Drug Administration (FDA) was olaparib (11)(12)(13). Evidence from phase I, II, and III studies demonstrated high clinical efficacy and excellent tolerability of olaparib therapy in BRCA1/2mutated OvCa patients (14)(15)(16)(17)(18)(19). However, it is noteworthy that only approximately 20% of OvCa patients have BRCA1/2 mutations (4,6,13). Defects in other HR-related genes, such as RAD51, RAD54, and PTEN genes, may also contribute to a "BRCAness profile" (4,20,21).
In the present study, we carried out a series of whole-exome sequencing (WES) on DNAs extracted from OvCa tissues of patients and identified the most frequently mutated genes in HR pathways. These highly mutated HR genes include BRCA1, BRCA2, RAD54B, FANCM, ATR, ATM, RAD51, and PALB2, and others. The regulatory mechanisms of these genes in DNA repair and the synthetic lethal effects when PARPi treatments were applied in various cancers with defective HR genes have been studied (22)(23)(24)(25)(26)(27)(28)(29)(30). RAD54B-deficient colorectal cancer cells were reported to increase accumulation of DNA DSBs and apoptosis under olaparib treatment (30). Its therapeutic value in OvCa remains unclear.
RAD54B was first isolated as a human RAD54 homolog in 1999 (31). Given that RAD54B interacts with RAD51 and DMC1 to enhance the D-loop formation and DNA-strand exchange during HR, this protein was suggested to be critical in DNA repair and harbored with tumor suppressor-like properties (32)(33)(34)(35). RAD54B was also found to form nuclear foci with hRAD54, hRAD51, and BRCA1 (36,37). Although more and more researchers have paid attention to the roles of RAD54B in anticancer therapy (38,39), the roles and molecular mechanisms of malfunctions of RAD54B in OvCa need further investigations. Here, we focused on RAD54B to explore whether RAD54B defects rendered OvCa cells more sensitive to a PARPi olaparib.
Interestingly, 7.3% of patients had RAD54B mutations (Fig. 1B). Four different RAD54B mutations were identified in six OvCa patients (Fig. 1C). Earlier studies have shown that RAD54B is an important player in HR pathway (31). However, its importance in anticancer therapy needs further investigations. In fact, alterations of RAD54B have been detected in several cancers, including lung adenocarcinoma, breast cancer, and hepatoma (44)(45)(46)(47). RAD54B silencing often leads to reduced cell proliferation and enhanced apoptosis (44,47). Moreover, in colorectal cancer cells, inactivation of RAD54B was reported to increase DSBs and apoptosis (37,48,49). Here, since 7.3% of OvCa patients harbored RAD54B mutations, the third highest mutated gene among these patients, we therefore focused on RAD54B to gain further insight into its therapeutic values in OvCas.

Pathogenicity prediction of RAD54B mutations
To predict the importance of RAD54B mutations in OvCa therapy, we first employed SIFT/PolyPhen2/Mutation assessor/Radial SVM software platforms. The clinical relevance of these variants was further analyzed and identified in COSMIC and ClinVar databases. The aforementioned WES analyses identified four RAD54B mutations, including N593S, H219Y, Y738H, and one stop-gained variant, in OvCa tissues of six patients (Fig. 1E). N593S was located at the conserved region between helicase motifs III and IV in SNF2 domain of RAD54B protein (Fig. 1F) (31). This mutation was observed in human primary lymphoma and colon cancer (31) and also found in acute myeloid leukemia (legacy mutation ID: COSM4416566) recorded in ClinVar database. To the best of our knowledge, the other two RAD54B missense mutations, H219Y and Y738H, were identified the first time in cancer cells. These two RAD54B missense mutations were not included in National Center for Biotechnology Information ClinVar or COSMIC databases. The RAD54B variants H219Y and Y738H were predicted to be damaging by SIFT platform and probably damaging by PolyPhen2 platform (Fig. 1E). In addition, a stop-gained mutation (CAG to TAG) at codon 209 was also identified. This stop-gained mutation was located at the N-terminal domain, which can lead to the loss of SNF2 domain and C-terminal domain in RAD54B protein (Fig. 1F). Loss of SNF2 domain can thus result in the disruption of ATPase activity, which was required for the translocation of RAD54B along duplex DNA and DNA double-helix opening (50). Collectively, these bioinformatic predictions suggested that these four RAD54B mutations can probably affect HR repair functions of RAD54B.

, A-D).
These results demonstrated that there were significantly more DSBs in RAD54B-mutated OvCa tissues when compared with the normal ovarian epithelial tissues.
Considering that RAD54B is an important protein involved in HR repair and taking into account the aforementioned prediction results via bioinformatics, we speculated that RAD54B mutations may affect its intracellular functions, thereby impairing DSB repair. To test this hypothesis, we conducted the following in vitro experiments.

RAD54B knockdown inhibited HR repair in OVCAR8 cells
To facilitate subsequent in vitro experiments, we first screened six OvCa cell lines preserved in our laboratory. Two cell lines, ES2 and OVCAR8, were selected for their abundant protein levels of RAD54B and PARP1 (Fig. 3A). To simulate the functional defects of RAD54B, its protein levels were knockdowned by using two different shRNAs (Fig. 3B). To test the aforementioned hypothesis, we used the DR-GFP assay, a commonly used method, to test intracellular HR repair abilities (56)(57)(58) in the following experiments (Fig. 3C). OVCAR8 cells were stably transfected with DR-GFP, and the assays were accomplished. The HR repair abilities of different groups were analyzed using flow cytometry, and the vector group without I-SceI transfection was used as the negative control. As shown in Figure 3, D and E, when I-SceI enzyme was transiently expressed, a portion of cells in the vector group gave GFP signals after HR, whereas almost none signals for the negative control group without I-SceI transfection. More importantly, when compared with the vector group, two groups with shRAD54B knockdown exhibited dramatically lower GFP-positive cells, indicating RAD54B knockdown inhibiting HR repair in OVCAR8 cells.

RAD54B knockdown enhanced sensitivity of OvCa cells to olaparib
To find out the potential influence of RAD54B malfunctions in OvCa therapy, we tested the effects of RAD54B knockdown on the sensitivities of OvCa cells to PARPi treatment. As a positive control, BRCA2 knockdown was used here (Fig. S1). Before we treated the cells with PARPis, the effects of RAD54B or BRCA2 knockdown on cell viabilities were first tested. The results showed that RAD54B knockdown, as well as BRCA2 knockdown, did not significantly suppress cell viabilities of ES2 and OVCAR8 cells (Fig. 4A). We then evaluated the effects of RAD54B knockdown on sensitivity of OvCa cells to olaparib. ES2 and OVCAR8 cells with or without RAD54B knockdown were treated with olaparib (11) (the first PARPi approved by FDA for the treatment of deleterious germline BRCA-mutated advanced OvCa) at different concentrations. As shown in Figure 4, B and C, RAD54B knockdown significantly increased sensitivities of ES2 and OVCAR8 cells to olaparib. The IC 50 values were decreased by 3.96-fold and 4.75-fold, respectively, via two shRNA constructs in ES2 cells. As for OVCAR8 cells, similar results were obtained (3.94-fold and 4.31-fold, respectively). As the positive control, BRCA2 knockdown increased the sensitivities of ES2 and OVCAR8 cells to olaparib by 3.32-fold and 3.13-fold, respectively (Fig. 4, B-D). These results demonstrated that RAD54B knockdown significantly enhanced sensitivities of OvCa cells to olaparib. RAD54B mutations enhance sensitivity of OvCa to PARPis

RAD54B knockdown increased DSBs in OvCa cells after olaparib treatment
It is widely accepted that when cells are treated with PAR-Pis, their single-strand DNA breaks will result in DSBs because they cannot be repaired in time. Cells with sound HR repair machineries might survive in this case, whereas the others might accumulate DSBs and eventually tend to die. Therefore, to investigate whether RAD54B knockdown can increase the accumulation of DSBs in olaparib-treated cells, we measured foci formation of γH2AX and 53BP1 in cells (Fig. 4, E and F). As shown in Figure 4E, both γH2AX and 53BP1 foci were significantly increased in RAD54B-knockdowned ES2 cells when compared with the vector group after they were treated with olaparib at the same concentration. Similarly, significant increases in percentages of 53BP1 and γH2AX foci-positive cells were also observed in RAD54B-knockdowned OVCAR8 cells compared with those of the control groups (Fig. 4F). In addition, Western blot results demonstrated increased protein levels of 53BP1 in RAD54B-knockdowned ES2 and OVCAR8 cells when they were treated with olaparib (Fig. 4G). All these data clearly demonstrated that RAD54B knockdown increased DSBs in OvCa cells after olaparib treatment. Considering that the cells of all groups are treated with olaparib under the same conditions, the generation of DSBs should thus be the same. Since the amounts of DSBs depend on the balance between generation and repair, therefore, the accumulation of DSBs implied a compromised HR repair caused by RAD54B knockdown.

RAD54B knockdown increased olaparib-induced apoptosis in OvCa cells
Next, we used TUNEL assays to detect whether RAD54B knockdown affects apoptosis in drug-treated cells. As shown in Figure 5, A and B, the results of TUNEL assays demonstrated that RAD54B knockdown dramatically increased apoptosis levels in olaparib-treated ES2 and OVCAR8 cells. Consistently, Western blot detection on antiapoptotic protein Bcl-2 and proapoptotic protein cleaved-PARP1 showed that the expression levels of Bcl-2 were decreased in RAD54B knockdown groups, whereas the expression levels of cleaved-PARP1 were elevated in olaparib-treated ES2 and OVCAR8 cells (Fig. 5, C  and D). The levels of cleaved-caspase 3, a ubiquitously distributed caspase and a main effector of the apoptotic cascades, were also significantly elevated in RAD54B-knockdown ES2 and OVCAR8 cells, when compared with the empty vector groups (Fig. 5, C and D). These results indicated that RAD54B knockdown dramatically increased apoptosis induced by olaparib treatment in OvCa cells.

RAD54B knockdown enhanced the inhibitory effects of olaparib on the growth of ES2 xenograft tumors
In the next experiments, ES2 cells (with or without RAD54B knockdown) were used to generate xenograft tumors (Fig. 6A). After olaparib treatment, the results showed that the weight and volumes of tumors in the RAD54B knockdown groups were significantly lower, compared with those in the vector group (Fig. 6, B-D).
To further delineate the molecular mechanisms of the inhibitory effects, the tumor tissues were subsequently analyzed by IHC. Protein levels of Ki67 were first examined in the xenograft tumor tissues. As shown in Figure 6E, the Ki67 levels were significantly decreased in RAD54B-knockdowned tumor groups. These results suggested that RAD54B knockdown significantly inhibited proliferation of OvCa cells with olaparib treatment. We then wondered whether RAD54B knockdown promoted apoptosis of olaparib-treated OvCa cells in xenograft tumors. As shown in Figure 6F, levels of cleavedcaspase 3 were significantly increased in the RAD54Bknockdowned groups. The RAD54B knockdown promoted olaparib-induced cell apoptosis in ovarian xenograft tumors. In addition, in situ TUNEL assays were carried out to study the DNA fragmentation status in the late stage of apoptosis. In Figure 6G, the significant increases of percentages of TUNELpositive cells in RAD54B-knockdowned groups indicated a higher level of apoptosis compared with that in the vector group. No significant difference was observed between the groups of shRAD54B-1 and that of shARD54B-3 on the levels of Ki67, cleaved-caspase 3, and TUNEL (Fig. 6, E-G). Altogether, these data suggested that RAD54B knockdown increased apoptosis and decreased cell proliferation in olaparib-treated ovarian xenograft tumors in vivo.

RAD54B mutations enhance sensitivity of OvCa to PARPis
Expression of wildtype RAD54B rather than RAD54B mutants abolished the effects of RAD54B knockdown In the aforementioned sections, we knockdowned RAD54B to simulate RAD54B malfunctions in cells and achieved the expected results. However, the more important question is whether RAD54B mutations we identified in patients indeed impair the normal functions of RAD54B. To answer this question, we conducted the rescue experiments by ectopically expressing different constructs of RAD54B in OvCa cells with RAD54B knockdown in advance. Toward this goal, we first knockdowned RAD54B in ES2 and OVCAR8 cells by using the shRNAs targeting 3 0 UTR of RAD54B mRNA (Figs. S2 and 7A). Subsequently, the RAD54B-knockdowned cells were complemented with different variants of RAD54B proteins, namely wildtype, N593S variant, or H219Y variant (Fig. 7A). As shown in Figure 7, B and C, the increased sensitivities of RAD54Bknockdowned cells to olaparib were brought back by expression of wildtype RAD54B rather than the two RAD54B mutants in both cell lines.
In addition, we analyzed the foci formation of γH2AX and 53BP1 to investigate whether RAD54B N593S or RAD54B H219Y expression alter the accumulation of DSBs in cells after olaparib treatment (Fig. 7D). As shown in Figure 7D, both γH2AX and 53BP1 foci were significantly increased in RAD54B-knockdowned ES2 and OVCAR8 cells (shRAD54B + control groups) when compared with their corresponding vector + control groups, reproducing the earlier results. At the same time, the expression of wildtype RAD54B significantly decreased the DSBs in RAD54B-knockdowned cells when we compared the results of the shRAD54B + wt group to those of shRAD54B + control group (Fig. 7D). However, the expression of RAD54B N593S (shRAD54B + RAD54B N593S group) or RAD54B H219Y (shRAD54B + RAD54B H219Y group) did not decrease the DSBs in RAD54B knockdown cells, compared with the results from shRAD54B + control groups in either cell line. These results indicate that expression of wildtype RAD54B can restore the functions after RAD54B knockdown, but expression of these two RAD54B mutants cannot restore its functions.
Western blot results in addition demonstrated that the increased levels of 53BP1 in RAD54B-knockdowned ES2 and OVCAR8 cells treated with olaparib were restored by expression of wildtype RAD54B rather than RAD54B N593S or RAD54B H219Y (Fig. 7E). More importantly, the results of DR-GFP assays showed that while expression of wildtype RAD54B restored the HR repair activities in RAD54Bknockdowned OVCAR8 cells after olaparib treatment, but both RAD54B N593S and RAD54B H219Y failed to do so (Fig. 7F).
Overall, our results demonstrated that expression of wildtype RAD54B rather than RAD54B mutants abolished the effects of RAD54B knockdown. These results therefore suggested that RAD54B N593S and RAD54B H219Y probably lost their functions in HR repair.

Expression of RAD54B in RAD54B-mutated tumors
Besides mutations, the expression levels of a protein may also affect its functions. We then measured the expression levels of RAD54B in RAD54B-mutated tumors using IHC. As shown in Fig. S3, the expression levels of RAD54B showed no significant differences in RAD54B-mutated tumors compared with those in the normal counterparts.

Discussion
DNA repair has been considered as a critical counteragent in carcinogenesis and potential targets for cancer therapy (59,60). Proteins implicated in HR repair pathway were frequently mutated in human cancers, such as BRCA1/2, ATM/ATR, RAD51, FA proteins, and others (4,59,60). Our WES analyses found that 22 HR genes were mutated in 44 OvCa cases out of a total of 82 cases (Fig. 1C). Among them, RAD54B has aroused our interest, because the mutation frequency of RAD54B was the third highest among 22 mutated HR genes (Fig. 1B). Since RAD54B plays an important role in HR repair, its mutation might impair DSB repair in cells. Consistent with this thought, cancer tissues harboring RAD54B mutations Figure 6. RAD54B knockdown enhanced the inhibitory effects of olaparib on the growth of ES2 xenograft tumors. A, schematic diagram depicting the ES2 xenograft mouse models. ES2 cells (with or without RAD54B knockdown) were subcutaneously injected into the right rib of each immune-deficient nude mice. The xenograft tumors were allowed to grow to 100 mm 3 . Olaparib was dissolved in a solution containing 30% castor seed oil and 4% DMSO before injection. The mice were intraperitoneally injected with olaparib at a dose of 150 mg/kg bodyweight every 2 days and then sacrificed 2 days after six administrations. The xenograft tumors were weighted and measured, and tumor tissues were collected for further studies. B, image of tumors after sacrifice. C, suppressive effects of RAD54B knockdown on the sizes of tumors. Tumor volumes were measured using a caliper. D, statistical analyses of the tumor weights. Mean ± SD, n = 5. *p < 0.05, **p < 0.01. E, Ki67 levels were decreased in RAD54B-knockdowned ES2 xenograft tumor cells with olaparib treatment. F, RAD54B knockdown increased the levels of cleaved-caspase 3 in RAD54B-knockdowned ES2 xenograft tumor cells with olaparib administration. G, effects of RAD54B knockdown on apoptosis by counting TUNEL-positive cells. Left panels, representative staining images of Ki67, cleaved-caspase 3, and TUNEL determined by IHC. Right panels, statistical analyses of the left panels. Mean ± SD, n = 3. **p < 0.01, ***p < 0.001. The scale bar represents 50 μm. DMSO, dimethyl sulfoxide; IHC, immunohistochemistry. contain more DSBs (Fig. 2). Considering that PARPis have superior effects in OvCas with HR deficiency (HRD) (61), we hypothesized that OvCa patients with RAD54B mutations prefer PARPi treatment. To test this hypothesis, we conducted a series of in vitro and in vivo experiments. Using DR-GFP assays, we demonstrated that RAD54B knockdown inhibited HR repair in OVCAR8 cells (Fig. 3). As expected, OvCa cells with RAD54B knockdown exhibited higher sensitivity to olaparib treatment (Fig. 4). Indeed, RAD54B knockdown induced accumulation of DSBs (Fig. 4) and aggravated apoptosis of OvCa cells induced by olaparib in vitro (Fig. 5). Olaparib inhibited the growth of xenograft tumors of ES2 cells with RAD54B knockdown more efficiently (Fig. 6). More importantly, when the endogenous RAD54B was replaced by the RAD54B mutants identified in patients, the OVCAR8 cells remained high sensitivities to olaparib treatment (Fig. 7). All these results suggest that inhibition of RAD54B functions results in defects in HR repair and increases sensitivities to olaparib in OvCa cells.
The earlier studies underestimated the existence of epithelial OvCas with HRD (3, 4, 62-66). Recent evidence has suggested that approximately 20 to 30% of epithelial OvCas harbor non-BRCA HRD (4). A couple of recent studies identified several frequently mutated HR genes in epithelial OvCas, including BRCA1/2, ATM, ATR, PTEN, RAD51, PALB2, CHEK2, FANCA/M, and others (63,64). The mutation frequencies of these HR genes assessed by targeted-next generation sequencing were close to those in our analyses by WES (Fig. 1B). However, RAD54B was neglected in these studies because of the limited list of the targeted gene panels. Notably, our WES probed into all the mutated HR genes in clinical epithelial OvCa specimens. RAD54B mutations were thus identified and aroused our interest. In our WES analyses, 7.3% of OvCa patients harbor RAD54B mutations, which is similar to the estimate of 6% in another study by Zhao et al. (65).
In the last decade, synthetic lethality led by PARPis in BRCA-deficient cancers has attracted considerable attention. The first PARPi, olaparib, was approved by FDA in maintenance therapy of BRCA1-or BRCA2-deficient OvCa patients in the year of 2014. Great efforts have been made toward the clinical development of PARPis afterward. In view of the superior effects, PARPis were approved for use in the first line The positive cells were determined by containing more than five 53BP1 or γH2AX foci. Mean ± SD, n = 3. #p > 0.05, **p < 0.01, or ***p < 0.001, compared with the vector-control group; ▫ p > 0.05, or § § p< 0.01 compared with the shRAD54Bcontrol group. E, Western blot analyses demonstrated the levels of 53BP1 in aforementioned cells. F, statistics of the percentages of GFP-positive cells in cells with indicated treatments. Mean ± SD, n = 3. #p > 0.05 or ***p < 0.001, compared with the vector-control group; ▫ p > 0.05 or § § § p < 0.001, compared with the shRAD54B-control group. DAPI, 4 0 ,6-diamidino-2-phenylindole. and early maintenance therapy along with the new data (67)(68)(69). In addition, scholars and oncologists have recently proposed a concept of HRD scoring for targeted and individualized therapy for OvCa patients (70)(71)(72). PARPi therapies are preferred for patients with higher HRD scores to achieve better outcome (71,72). HRD scoring is heavily dependent on HRD gene panel, in which RAD54B is not currently included. Our study reveals that RAD54B is a frequently mutated HR gene in OvCa patients, and its malfunction enhances the sensitivity of cells to olaparib. This evidence supports RAD54B as a novel biomarker that should be included in HRD gene panel to benefit more patients.
Overall, our study provides evidence for potential applications of PARPis in synthetic lethal therapy and potential individualized therapy for patients with RAD54B mutations.

Clinical specimens
The study was approved by the Ethics Committee at Jiangsu University. The studies in this work abide by the Declaration of Helsinki principles. The formalin-fixed paraffin-embedded (FFPE) OvCa tissue blocks from 2007 to 2018 were collected at the Affiliated Hospital of Jiangsu University. All the FFPE OvCa and normal tissues were dissected from surgical specimens. Diagnosis was confirmed by pathologists. Major clinical characteristics of these patients are listed in Table S1. In addition, six normal ovarian epithelial tissue samples were collected as negative control.

WES and variant confirmation
The FFPE OvCa tissue blocks were deparaffinized with xylene. Genomic DNAs were extracted by using QIAamp DNA FFPE Tissue Kit (Qiagen) and randomly fragmented by Corvatis Technology. The selected DNA fragments (distributed between 150 and 250 bp) were dA-tailed. Adapters were ligated to both ends of the DNA fragments. The adapterligated DNA fragments were amplified by ligation-mediated PCR, purified, hybridized to human exome array for enrichment, and followed by circularization. The DNA nanoballs were produced by rolling circle amplification. The exome libraries were sequenced by BGISEQ-500 sequencing platform (BGI). Data were processed by BGISEQ-500 basecalling software and stored in FASTQ format.
The FASTQ raw data were filtered and aligned to human reference genome (GRCh37/HG19) using Burrows-Wheeler Aligner software. Variant callings were performed by the GATK-recommended variant analysis process (GATK, https:// www.broadinstitute.org/gatk/guide/best-practices). Duplicate reads were removed, and coverage metrics were calculated by Picard-tools (version 1.48; http://picard.sourceforge.net/). Local insertion-deletion realignment and base quality score recalibration were conducted by using GATK. The nonpathogenic variants with mutation frequency >1% in 1000 human genome project database, NHLBI-ESP6500 European American database, and NHLBI-ESP6500 African American database were further removed. The clinical significance and status (including pathogenic/nonpathogenic and tolerated/ possible or probably damaging) of variants were identified in COSMIC and ClinVar databases. The potential pathogenic variants were in addition analyzed by using SIFT/PolyPhen2/ Mutation assessor/Radial SVM software platforms. See Tables S2 and S3 for a full list of the WES-identified HR mutations.

IHC
IHC assays were performed as previously described (73)(74)(75). Briefly, the tumor tissues dissected from mice and fixed in neutral-buffered formaldehyde for 3 h at 25 C. Tissues were dehydrated in 70, 85, 95, and 100% ethanol and incubated in xylene. Subsequently, the tissues were embedded in paraffin, and paraffin-embedded tissues were sliced at 5 μm.
The paraffin-embedded tissue slices were dewaxed followed by antigen retrieval. The slices were then incubated with relevant primary antibodies and incubated with corresponding horseradish peroxidase-conjugated secondary antibodies for 20 min at 25 C, followed by visualization with 3,3diaminobenzidine, and then counterstained with hematoxylin. After dehydration and sealing, immunohistochemically stained proteins were observed in situ using Nikon Eclipse. Experiments were independently performed three times. The information of antibodies is listed in Table S4.

Cell culture and reagents
OvCa ES2 and OVCAR8 were cultured in RPMI1640, and HO8910 cells were cultured in Dulbecco's modified Eagle's medium, supplemented with 10% fetal bovine serum, and 100 U/ml penicillin and 100 μg/ml streptomycin (Thermo Fisher Scientific) at 37 C in an atmosphere containing 5% CO 2 , respectively. Olaparib was purchased from Selleck Chemicals and diluted in dimethylsulfoxide. All other chemicals were purchased from Sigma-Aldrich unless otherwise specified.

Western blotting
Western blotting assays were performed as previously described (73)(74)(75). Briefly, cells were lysed with ice-cold radioimmunoprecipitation assay lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM PMSF, and 1% Triton X-100), and the concentrations of the extracted proteins in supernatant were quantified by bicinchoninic acid protein assay kit (Beyotime Biotech). About 30 μg of each protein sample was loaded on SDS-PAGE gel. After the SDS-PAGE electrophoresis, proteins were transferred on a polyvinylidene difluoride membrane, and blocked with 5% skim milk in Tris-buffered saline with Tween-20 (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.05% Tween-20). Specific proteins were probed with the relevant primary antibodies followed by secondary antibodies. The immunoblots were visualized with ECL plus Chemiluminescence kit (Thermo Fisher Scientific). The antibodies used are listed in Table S4.

Lentivirus infections
The shRNAs were purchased from BGI. The sequences are listed in Table S6. Lentivirus was packaged by following the manufacturer's instructions of a ViraPower Kit from Thermo Fisher Scientific. Production and infection of lentiviruses were conducted as described previously (73). Cells infected with the lentiviruses were screened by 1 μg/ml puromycin.

HR assays
Shown as in Figure 3C, OVCAR-8 cells were transfected with linearized DR-GFP (modified from Addgene #26475) reporter constructs using Etta X-Porator H1 (Etta Biotech) electroporator. G418 was used to select the stable clones. Stable cell lines containing the DR-GFP reporter constructs were infected with lentivirus raised from shRAD54B-1/3 plasmids or vectors and were selected with 1 μg/ml of puromycin. Then, cells were transfected via electroporation at a final amount of 1 μg of pCBASceI (Addgene #26477). After 48 h, GFP-positive cells were counted using a Gallios Flow Cytometer (Beckman Coulter). Data were analyzed using FlowJo software (BD Biosciences).

Immunofluorescence staining
Immunofluorescence staining was carried out as described previously (75,76). Briefly, cells (3 × 10 4 cells per well/24-well plate) were fixed in 4% paraformaldehyde for 10 min at 25 C. Fixed cells were then permeablized with 0.5% Triton X-100 for 5 min at 25 C and blocked with 3% bovine serum albumin in PBS for 30 min. Next, the cells fixed on a slide were incubated with a selected primary antibody at an appropriate dilution for 2 h at 25 C in the dark, followed by incubating with an appropriate fluorescent secondary antibody for 1 h at 25 C. The slides were stained with 1 μg/μl of 4 0 ,6-diamidino-2phenylindole for 5 min at 25 C and observed under Nikon Eclipse. The antibodies are described in Table S4.

TUNEL assays
The paraffin-embedded tissues from mice were fixed in 4% paraformaldehyde solution for 30 min at room temperature. The tissues on slices were then washed with PBS and incubated in 0.5% Triton X-100 for 7 min. The TUNEL assay was carried out by following the manufacturer's instructions of the TUNEL BrightGreen Apoptosis Detection kit from Vazyme. TUNEL-stained cells were observed in situ using Nikon Eclipse.

In vivo xenograft tumor model
All procedures involving mice were approved by the Institutional Animal Care and Use Committee of Jiangsu University. Six-week-old nude BALB/c mice were purchased from Yangzhou University, bred, and maintained in a pathogen-free facility. For xenograft mouse models, 0.2 ml of 1 × 10 7 of RAD54B-knockdowned ES2 cells was subcutaneously injected into the right rib of each mice. When the tumors grew to about 100 mm 3 , tumor-bearing mice were randomly divided into three groups with five mice per group. The mice were intraperitoneally administrated with olaparib at a dose of 150 mg/ kg bodyweight every 2 days. Two days after six administrations, mice were sacrificed, and the xenograft tumors were weighted. Tumor tissues were collected for further studies.

Statistical analysis
Data are presented as mean ± SD unless otherwise stated. Statistical significance between groups was assessed by Student's t test (two-tailed) or one-way ANOVA using GraphPad Prism, version 7.00. In IHC analyses, p values between groups were calculated by Chi-square test (n > 40) or Fisher's exact test (n < 40). Statistical significance with p < 0.05 was considered significant. #p ≥ 0.05; *p < 0.05; **p < 0.01; and ***p < 0.001.

Data availability
All data are contained within the article.
Supporting information-This article contains supporting information.