Therapeutic Potential of the Poly(ADP-ribose) Polymerase Inhibitor Rucaparib for the Treatment of Sporadic Human Ovarian Cancer Free

Ovarian cancer is the second most common gynecologic malignancy in the United States (1). Despite radical surgery and initial high response rates to platinum- and taxane-based chemotherapy, most patients experience a relapse, with a median progression-free survival of only 18 months (2). Thus, advances in the understanding of the molecular pathogenesis of ovarian cancer coupled with the development of novel-targeted therapies are needed to improve outcomes. Patients with ovarian cancer with germline mutations in either BRCA1 or BRCA2 genes exhibit impaired ability to repair DNA double-strand breaks (DSB) via homologous recombination, and show a heightened sensitivity to inhibitors of a second DNA repair pathway, the base excision repair (BER) pathway (3). PARP is a nuclear protein that senses and binds to DNA single-strand breaks (SSB) and subsequently activates the BER pathway by recruiting additional repair factors (4). When PARP is inhibited, persistent SSBs become DNA DSB during DNA synthesis via collapsed replication forks (5, 6). To repair DSBs, the cell preferentially uses homologous recombination, which is usually considered error proof. If the cell cannot initiate homologous recombination, as is the case with BRCA1/2-mutant tumors, it resorts to more error-prone pathways, such as nonhomologous end joining or single-strand annealing, which can cause gross chromosomal mutations, growth inhibition, and eventual cell death (3).

Clinical studies have confirmed the activity of PARP inhibitors in patients with ovarian cancer with germline BRCA1/2 mutations (7, 8). However, recent clinical data indicate that a subset of patients with sporadic ovarian cancer (with wild-type BRCA1/2) may also respond to PARP inhibition, suggesting that BRCA1/2 mutations may not be the sole predictors of response (8, 9). These clinical findings clearly support the hypothesis that PARP inhibitors may also be effective in ovarian cancers bearing other deficiencies in homologous recombination, which may occur in a high proportion of sporadic epithelial ovarian cancer cases (10). In this context, compromised activity of ATM or other proteins involved in homologous recombination (11), overexpression of AURKA (12), or loss of PTEN function (13), has been described to lead to enhanced sensitivity to PARP inhibitors. Furthermore, it has been proposed that amplification of EMSY, which is capable of silencing BRCA2, may also lead to enhanced sensitivity to PARP inhibitors (14). However, the role of these markers in predicting response to PARP inhibitors remains controversial and preclinical studies are limited (15, 16).

Rucaparib (CO-338, formerly known as AG014699 and PF-01367338) is a highly selective inhibitor of PARP1 and PARP2 proteins (with an inhibition constant of <5 nmol/L), which has recently shown oral bioavailability (17). An earlier report shows antiproliferative activity of rucaparib in breast and pancreas cancer cell lines as well as an immortalized ovarian surface epithelial cell line each harboring either methylation or mutations of BRCA1/2 (18). Here, we show that rucaparib may also be useful for the treatment of sporadic ovarian cancer lacking BRCA1/2 silencing through methylation or mutations. Moreover, rucaparib was able to potentiate the cytotoxicity of DNA-damaging chemotherapeutic agents independent of its activity as single agent. To show this, we tested the in vitro effects of rucaparib against a panel of 39 ovarian cancer cell lines, representing all histologic subtypes of the disease. We then sought to validate known and identify novel response markers. For this purpose, cell lines were characterized for BRCA1/2 promoter methylation and mutational status. Other potential response predictors such as those known to be directly implicated in DNA repair, but also PTEN, AURKA, and EMSY were studied using gene expression profiling, Western blot analysis, and array comparative genomic hybridization (CGH). Cells were also characterized for their sensitivity to platinum-based chemotherapy. Multiple drug effect/combination index isobologram analysis was conducted to study drug interactions between rucaparib and chemotherapeutic agents commonly used for the treatment of ovarian cancer. This was done both using cell lines that were either sensitive or resistant to single-agent rucaparib. To better understand observed synergies, we finally studied the effect of rucaparib on chemotherapy-induced apoptosis, DNA fragmentation, and γH2AX formation.

The effects of rucaparib and chemotherapeutics on growth inhibition were studied in a panel of 39 established human ovarian cancer cell lines. Individuality of each cell line was checked by mitochondrial DNA sequencing once obtained from the source. Cell lines were passaged for less than 3 months after authentication. The cell lines were obtained from American Type Culture Collection (CAOV3, ES-2, OV90, OVCAR3, TOV112D, TOV21G), the European Collection of Cell Cultures (A2780, OAW28, OAW42), the German Tissue Repository, DSMZ (COLO704, EFO21, EFO27), Dr. Viji Shridhar, Mayo Clinic, (Rochester, MN; DOV13, HEYC2, OV167, OV177, OV207, OVCAR5), Dr. David T. Curiel, University of Alabama at Birmingham [Birmingham, AL (HEY)], Dr. Hiroaki Itamochi, Department of Obstetrics and Gynecology, Tottori University School of Medicine (Tottori, Japan; KK and KOC-7c), Dr. Beth Karlan, Cedars Sinai (Los Angeles, CA; OV2008, OVCA420, OVCA429, OVCA432, SKOV3), the Japanese Health Science Research Resources Bank, (Osaka, Japan; KURAMOCHI, MCAS, OVISE, OVKATE, OVMANA, OVSAHO, OVTOKO, RMG1, RMUGS, TYK-nu), and Dr. Simon P. Langdon, Edinburgh Cancer Research Center, University of Edinburgh (Edinburgh, United Kingdom; PE014, PEA2, PEO6). Detailed information on culture media and reagents is provided in the Supplementary Materials and Methods (Supplemental Table S1).

Cells were plated into 24-well tissue culture plates at a density of 5 to 20 × 10³ and grown with or without increasing concentrations of rucaparib (ranging between 10 μmol/L and 0.005 μmol/L). Cells were counted on day 1 and 6 using a Coulter Z2 particle counter (Beckman Coulter Inc.). Growth inhibition was calculated as a function of the number of generations inhibited in the presence of rucaparib versus the number of generations over the same time course in the absence of the drug. To study the inhibition of anchorage-independent growth, soft agar assays were conducted. A 0.5% agar solution (Difco Agar Noble, BD) was placed on the bottom of a 24-well plate. Cells were seeded in quadruplicates of 5 × 10³ and mixed into the 0.3% agar top layer that had been prepared with or without 1 μmol/L rucaparib. Culture plates were stored at 37°C, 5% CO₂ for up to 5 weeks. Colonies were stained with Neutral Red solution (Sigma-Aldrich). All assays were conducted at least 3 times in duplicate for each cell line.

Sanger sequencing was used to screen the entire BRCA1/2 gene for mutations. The coding region of BRCA1 and BRCA2 were amplified by PCR. The pairs of primer sequences used have been reported previously (19). Sequencing was carried out as previously described by Beckman Coulter Genomics (20). Mutation surveyor software (SoftGenetics) was used to visually analyze sequencing traces for mutations and all potential variants were confirmed by an independent PCR and sequencing reaction. In addition, all cell lines with initially detected BRCA1/2 variants were reordered from the original supplier and authenticated by short tandem repeat sequencing in addition to the preceding mitochondrial DNA sequencing. Mutations were then confirmed in these cell lines by an independent third PCR and sequencing reaction. Missense variants were analyzed for their functional significance using the 2 programs SIFT (21) and Polyphen (22). Accession number used for BRCA1 is NM_007294.3 and for BRCA2 is NM_000059.3.

BRCA1 and BRCA2 promoter methylation was measured by bisulfite PCR and pyrosequencing. Assays are described in detail in the Supplementary Materials and Methods. CpG islands examined are shown in Supplementary Fig. S1.

Microarray hybridisations of 39 ovarian cell lines were conducted using the Agilent Human genome 44K array chip, as described previously (23). The original data are available online with the Gene Expression Omnibus accession number GSE26805.

Genomic DNA was extracted from frozen cells using the DNeasy Blood and Tissue Kit (Qiagen). Labeling, hybridisation, and analysis of Agilent 105K oligonucleotide CGH arrays were conducted according to the manufacturer's protocol (Human Genome CGH 105A Oligo Microarray Kit, version 5.0, Agilent Technologies) and have been described previously (24). Log₂ ratios more than 1 were considered to be amplified (2-fold increase).

The relevant exons of the PTEN gene in each cell line were PCR amplified, sequenced, and assessed for potential sequence alterations using approaches previously described (20). The nucleotide sequences were analyzed using the Mutation Surveyor program (Soft Genetics LLC) and through visual inspection. All somatic mutations were confirmed by independent PCR and sequencing reactions.

Western blot analysis for PTEN was conducted as previously described (25). PTEN expression was detected, using a monoclonal anti-PTEN antibody (PTEN 7974; Santa Cruz Biotechnology).

Multiple drug effect analysis using rucaparib in combination with carboplatin, doxorubicin, topotecan, paclitaxel, and gemcitabine was conducted as described previously (26). Combination index values were derived from variables of the median effect plots, and statistical tests were applied to determine whether the mean combination index values at multiple effect levels (IC₂₀–IC₈₀) were significantly different from combination index = 1. In this analysis, synergy is defined as combination index values significantly lower than 1.0, antagonism as combination index values significantly higher than 1.0, and additivity as combination index values equal to 1.0.

Effects of rucaparib on apoptosis were conducted using an Annexin V-FITC Apoptosis Detection Kit (MBL) and flow cytometry. Cells were exposed to 5 μmol/L of rucaparib alone or in combination with 10 nmol/L topotecan. After 3 days, samples were analyzed using the Cell Lab Quanta SC flow cytometer (Beckman Coulter Inc.).

To detect single- and double-strand DNA breaks in cell lines exposed to rucaparib alone or in combination with topotecan a comet assay was conducted in accordance to the manufacturer's instructions (Trevigen). In brief, control and treated cells were mixed with LMagarose and spread on to comet slides. The comet slides were immersed in the lysis solution and incubated for 60 minutes at 4°C. Following lysis the slides were immersed in freshly prepared alkaline buffer for 60 minutes in the dark. Electrophoresis was conducted at a set voltage (21V for 30 minutes), dehydrated in 70% ethanol, and stained with DNA-bound SYBR Green I fluorescence stain. For visualization of DNA damage images were captured using a confocal microscope at 494/521 nm wavelength.

HEY, MCAS, and MDA-MB436 (BRCA1 deficient breast cancer cell line) were seeded on sterile cover slips in 12-well plates and treated with 3μmol/L rucaparib or 30 nmol/L topotecan or the combination of both. After 24 hours, cells were washed with PBS, fixed with 4% paraformaldehyde for 20 minutes, and permeabilized with 70% ethanol for 5 minutes at −20°C. Cells were stained with the primary antibody against Rad51 (H-92, Santa Cruz Biotechnology) or γH2AX (Ser139, Cell Signaling) and incubated for 1 hour with the secondary antibody Alexa 488 or Alexa 594 (Invitrogen) for Rad51 or γH2AX, respectively. After washing with PBS, a drop of Prolong Gold antifade reagent with 4,6-diamidino-2-phenylindole Invitrogen) was applied for counterstaining. Images from random fields were taken using a Nikon Eclipse E 800 microscope with an attached camera and using Spot Software from Diagnostic Instruments (Diagnostic Instruments).

Associations between biomarkers and in vitro sensitivity were analyzed using Spearman rho correlation, and a bootstrapping procedure was implemented to improve statistical robustness. In addition a Resolver system ANOVA was conducted on the ovarian cancer cell lines classified by response to rucaparib across a DNA repair gene set of 683 sequences on the Agilent Whole Human Genome platform classified as “DNA damage and repair.” We used a statistical value for sequences of 1.75 change in at least 3 experiments and a P < 0.05. Differences between subgroups were compared using the Student t test, χ² test, and Mann–Whitney U test. To analyze which chemotherapeutic agent was potentiated the most by rucaparib and at which chemotherapy concentration the greatest effect was seen, a multivariate ANOVA model was conducted. Interaction terms were included to model differences between cell lines, chemotherapeutic agents, and specific drug concentrations. All reported P values are 2 sided, and statistical significance was reached at P values less than 0.05.

The in vitro effects of rucaparib on human ovarian cancer cells were evaluated using a panel of 39 established human ovarian cancer cell lines. These cells lines were selected to be representative of a range of histologic ovarian cancer subtypes (Table 1). The effective dose range (IC₂₀–IC₈₀) was identified using a wide range of rucaparib concentrations (10–0.005 μmol/L). The IC₅₀ values varied significantly between individual cell lines and ranged from 2.5 μmol/L for COLO704 to more than 15 μmol/L in 13 of the 39 investigated cell lines (Table 1 and Fig. 1). There was no statistically significant correlation between the histologic subtype and sensitivity to rucaparib (data not shown). Next, we studied the effect of rucaparib on anchorage-independent growth using soft agar assays. Of the 39 tested cell lines, 22 ovarian cancer cell lines formed colonies in soft agar (Supplementary Fig. S2). The inhibition of colony formation varied significantly between individual cell lines when treated with 1 μmol/L rucaparib and ranged between 80% in sensitive lines and no growth inhibition in resistant cell lines. Importantly, the observed response toward rucaparib in the 3-dimensional soft agar experiments correlated well with that seen in the 2-dimensional anchorage-dependent growth assay (accuracy 82%, P = 0.009).

Preclinical and clinical evidence indicates that ovarian cancer cells with BRCA1/2 mutations are sensitive toward PARP inhibitors (6, 7, 27). To better understand the predictive role of BRCA1/2 mutations in ovarian cancer, we sequenced the BRCA1 and BRCA2 genes in each of the 39 cell lines. Five ovarian cancer cell lines showed BRCA1/2 gene variations (Table 2). A deleterious BRCA2 nonsense mutation (c.6952C>T, p.R2318X) was identified in the cell line KURAMOCHI and 2 BRCA1 missense mutations were found in the cell line KOC-7c (c.39T>A, p.N13K and c.395A>C, p.N132T). Furthermore, BRCA2 missense mutations were found in the cell lines MCAS (c.964A>C, p.K322Q), OVCA420 (c.6131G>A,pG2044D), and OVMANA (c.2275C>G, p.L759V). Because it is difficult to interpret the functional significance of missense mutations, we further characterized the predicted functional significance of these using both prediction software programs SIFT and Polyphen (21, 22). The BRCA1 missense mutations in KOC-7c cells were predicted to be damaging by both programs. The BRCA2 variants in OVMANA and OVCA420 cells were deemed to be benign using both the programs, and the third variant (c.964A>C, p.K322Q) in MCAS cells was considered to be damaging by SIFT and benign by Polyphen. In summary, of the 2 human ovarian cancer cell lines with deleterious BRCA1/2 mutations, one line was sensitive (KURAMOCHI, IC50 <5 μmol/L) and one was resistant (KOC-7c, IC₅₀ >15 μmol/L) to treatment with single-agent rucaparib. These findings confirm the clinically observed activity of PARP inhibitors in hereditary ovarian cancer caused by germline mutations in BRCA1/2, but also suggest that the presence of a mutation may not suffice to ensure activity of a PARP inhibitor. Most importantly, however, these correlative studies clearly indicate that multiple cell lines with wild-type BRCA1/2 showed in vitro sensitivity comparable with that seen in the BRCA2-mutated KURAMOCHI cell line (Fig. 1).

A recent study found the breast cancer cell line UACC3199, which shows epigenetic silencing of BRCA1, to be sensitive to rucaparib (18). To elucidate whether BRCA1/2 promoter methylation would be associated with response to rucaparib in human ovarian cancer cell lines, we sought to quantify BRCA1 and BRCA2 CpG island methylation in each of the cell lines. This was done by bisulfite pyrosequencing of CpG sites relevant to transcription control (Supplementary Fig. S1). Importantly, we did not find high-level methylation of BRCA1 or BRCA2 promoter regions in any of the 39 examined ovarian cancer cell lines. The highest average methylation level of 10 CpG islands in the BRCA1 promoter region was seen in the RMG1 cell line (average 18%, maximum 34% in CpG7). The highest average methylation level of 7 CpG islands in the BRCA2 promoter region was seen in OVISE cells (average 17%, maximum 31% in CpG7; data not shown).

Preclinical and preliminary clinical evidence suggests that loss of other proteins involved in DNA DSB repair other than BRCA1/2 may be associated with sensitivity to PARP inhibition (8, 9, 11). Gene expression profiles for the cell line panel were generated using the Agilent 44K chips. A correlation analysis between gene expression levels and rucaparib in vitro sensitivity was restricted to 683 genes involved in DNA damage and repair, as defined by the Gene Ontology Annotation database (http://www.ebi.ac.uk/goa). Of these, 71 genes were significantly correlated with the IC₅₀ values for rucaparib (P < 0.05; Table 3). As such, cell lines with low expression of ATM, RAD51L1, RAD50, MSH5, or MLH1 showed lower IC₅₀ values when compared with cell lines with higher gene expression (ATM, r = 0.42, P = 0.009; RAD51L1, r = 0.38, P = 0.018; RAD50, r = 0.320, P = 0.049; MSH5, r = 0.349, P = 0.032; and MLH1, r = 0.325, P = 0.046; Table 3 and Supplementary Fig. S3). In addition, low expression of BCCIP (BRCA2- and CDKN1A-interacting protein) or BRCA1–BRCA2-containing complex, subunit 3 (BRCC3) was significantly correlated with sensitivity toward rucaparib (r = 0.47, P = 0.003, r = 0.34, P = 0,036), respectively. By conducting ANOVA classified by response to rucaparib (sensitive cell lines, IC₅₀ < 10 μmol/L; resistant cell lines, IC₅₀ > 10 μmol/L) across the 683 genes involved in DNA damage and repair, a set of 57 differentially-expressed genes was identified confirming the significance of the majority of the response predictors that had been initially identified using a Spearman correlation (Table 3).

Next, we sought to study additional biomarkers that have previously been suggested to predict response to PARP inhibitors. Amplification of AURKA was associated with low IC₅₀ values (IC₅₀ < 5μmol/L, χ² test, P = 0.023; Table 1). Similarly, amplification of EMSY was associated with low IC₅₀ values (IC₅₀ < 5 μmol/L, χ² test, P = 0.023; Table 1). PTEN mutations were found in 5 of 39 ovarian cancer cell lines, and each was associated with low-protein expression. However, PTEN mutations were not statistically significantly associated with low IC₅₀ values (IC₅₀ < 5 μmol/L, χ² test, P = 0.530; IC₅₀ < 10 μmol/L, χ² test, P = 0.768; Table 1).

Recent clinical data in recurrent ovarian cancer suggest that patients with platinum-sensitive recurrent sporadic ovarian cancer may particularly benefit from single-agent PARP inhibitor treatment or maintenance therapy (8, 9). To validate whether preclinical platinum sensitivity is associated with response to rucaparib, we determined the IC₅₀ concentrations for carboplatin in each cell line. Statistical analysis comparing the growth-inhibitory effects of rucaparib (IC₅₀s) with the growth-inhibitory effects of carboplatin reveals a statistically significant positive correlation between in vitro sensitivity to rucaparib and carboplatin (r = 0.61, P < 0.001; Table 1) confirming that the assessment of platinum sensitivity in ovarian cancer may be a clinically useful enrichment strategy for selection of patients most likely to respond to PARP inhibitor therapy (8, 9). A recent chemical compound profiling study using differential scanning fluorimetry showed that the 2 PARP inhibitors rucaparib and olaparib have comparable binding selectivity for the catalytic domains of PARP 1, 2, 3, and 4 (28). In vitro rucaparib and olaparib showed comparable efficacy across the present ovarian cancer cell line panel [mean IC₅₀ (range) for rucaparib was 9.4 μmol/L (2.5 >15 μmol/L) and for olaparib was 10.4 μmol/L (1.2 >15 μmol/L), r = 0.68, P < 0.001].

Multiple drug effect analyses were conducted to determine the nature of interactions between rucaparib and 5 chemotherapeutic agents most commonly used for the treatment of ovarian cancer: carboplatin, paclitaxel, doxorubicin, topotecan, and gemcitabine (Supplementary Fig. S4). These studies were conducted using 5 ovarian cancer cell lines with varying degrees of sensitivity toward single-agent rucaparib with IC₅₀ values ranging between 3.9 μmol/L in A2780 and more than 15 μmol/L in MCAS or OV2008 cells. Synergistic interactions were observed when rucaparib was combined with topotecan in all of the 5 cell lines examined [mean combination index values ranged between 0.53 [95% confidence interval (CI), 0.36–0.70, P < 0.001] and 0.72 (95% CI, 0.59–0.85, P < 0.001); Fig. 2A]. Importantly, synergistic drug interactions with topotecan were seen in rucaparib-sensitive cells (A2780, KK) but also in cell lines that were less (HEY) or not sensitive to single-agent rucaparib (MCAS and OV2008; Fig. 2B). Synergistic interactions were also observed when rucaparib was combined with doxorubicin in 4 of 5 lines [mean combination index values ranged between 0.69 (95% CI, 0.56–0.82, P < 0.001) and 0.82 (95% CI, 0.49–1.15, P = 0.500); Fig. 2A]. Synergistic and additive interactions were observed when rucaparib was combined with carboplatin [mean combination index values ranged between 0.45 (95% CI, 0.22–0.68, P = 0.001) and 1.07 (95% CI, 0.94–1.20, P = 0.240)] and paclitaxel [mean combination index values ranged between 0.73 (95% CI, 0.59–0.85, P < 0.001) and 1.00 (95% CI, 0.65–1.35, P = 0.980)]. In contrast, additive interactions were observed when rucaparib was combined with gemcitabine in each of the 5 lines examined [mean combination index values ranged between 0.98 (95% CI, 0.81–1.15, P = 0.820) and 1.22 (95% CI, 0.95–1.49, P = 0.104); Fig. 2A]. By using a multivariate ANOVA model (accounting for type of chemotherapy, cell line, and drug concentration as single factors), we analyzed the benefit of adding rucaparib to chemotherapy in more detail. Using this multivariate approach, we were able to show that the drug interactions differed significantly between individual chemotherapeutic agents (P = 0.010). When adding rucaparib to chemotherapy, the benefit was most pronounced for topotecan, less so for doxorubicin or carboplatin, and the least for paclitaxel or gemcitabine (Supplementary Fig. S5). Moreover, when combining rucaparib with topotecan, synergistic interactions were observed across the entire concentration range (IC₂₀–IC₈₀). Importantly, however, the largest increase in rucaparib-induced enhanced growth inhibition was seen at lower topotecan concentrations (P = 0.018; Supplementary Fig. S6). We believe this is an important observation as data from a recent clinical trial show that due to an enhanced myelosuppression dose reductions are necessary in the clinical setting when combining topotecan with the PARP inhibitor olaparib (29). Nonetheless, these preclinical data suggest that significant antitumor activity may still be achievable despite a clinically necessary reduction in the dose of chemotherapy.

Synergistic drug interactions were not restricted to rucaparib-sensitive ovarian cancer cell lines but were also seen in cell lines considered to be less sensitive or resistant to single-agent rucaparib such as HEY (IC₅₀ > 10 μmol/L) and MCAS (IC₅₀ > 15μmol/LM) cells. To better understand these synergistic interactions observed in lines with low sensitivity or even resistance toward rucaparib as single agent, we compared the effects of rucaparib, topotecan, and the combination of both on apoptosis, DNA fragmentation, and the formation of DNA DSBs using the HEY and MCAS cells known to be resistant to PARP inhibition as a single agent. Rucaparib alone did not induce apoptosis in these cell lines. However, adding rucaparib to topotecan did lead to an increase in the fraction of cells undergoing apoptosis when compared with treatment with topotecan alone (Fig. 3A). Next, we conducted single-cell gel electrophoresis to study DNA fragmentation, which can be visualized by the size of a cell's comet tail in the depicted fluorescent electrophoresis image (Fig. 3B). Rucaparib alone only mildly increased DNA fragmentation, however, adding rucaparib to topotecan drastically increased DNA fragmentation when compared with topotecan alone (Fig. 3B). Subsequently, we investigated the effects of rucaparib and topotecan on the formation of DNA DSBs by staining for γH2AX formation, which accumulates and becomes phosphorylated at sites of broken DNA DSBs, and for RAD51 expression, which aids in DNA DSB repair and serves as a marker of the cells ability to repair DNA DSBs. Importantly, γH2AX foci formation, a marker of DNA DSBs, only slightly increased following treatment with rucaparib alone, and markedly increased following treatment with topotecan alone. Nonetheless, γH2AX foci formation seemed to be most pronounced when both drugs were combined (Fig. 4). In summary, despite a relative insensitivity of HEY and MCAS ovarian cancer cells to single-agent rucaparib, treatment with the respective PARP inhibitor did seem to enhance topotecan-induced apoptosis, DNA fragmentation, and γH2AX formation. These results are consistent with the observed synergy between rucaparib and topotecan in HEY and MCAS cells and support the premise that PARP inhibitors may potentiate the cytotoxicity of DNA-damaging agents independent of their activity as a single agent.

Clinical proof-of-principal synthetic lethality with PARP inhibitors has been shown in hereditary ovarian cancer caused by germline mutations in BRCA1/2 (7). However, the potential of these drugs for the treatment of ovarian cancer beyond BRCA mutation carriers has yet to be defined. Results of 2 recent clinical studies suggest that patients with recurrent ovarian cancer without BRCA1/2 germline mutations may benefit from treatment with a single-agent PARP inhibitor (8, 9). Consistent, with these clinical observations, we found activity of the potent PARP inhibitor rucaparib not to be restricted to ovarian cancer cells harboring BRCA1/2 mutations. These findings may be explained by the results of a recently published comprehensive molecular characterization of high-grade serous papillary ovarian cancer, which identified multiple aberrations in genes involved in DNA DSB repair in addition to those seen in BRCA1/2 (10). Of 316 patients with high-grade serous papillary ovarian cancer, 154 (49%) had aberrations in genes that may affect homologous recombination. Although 20% of the patients had germline or somatic BRCA1/2 mutations, an additional 11%, 6%, and 5% of the patients had epigenetic silencing of BRCA1, amplification of EMSY, or loss of PTEN expression, respectively. Moreover, an additional 7% of the patients had mutations in RAD51C, ATM, ATR, or Fanconi anemia genes (10). It is, however, currently not clear whether these aberrations all lead to a decrease in homologous recombination sufficient enough to cause sensitization of tumor cells toward PARP inhibitors. Indeed, the results of our preclinical studies support the hypothesis that some of these aberrations may be synthetically lethal when combined with PARP inhibition. In line with earlier studies, we are able to confirm that low expression of ATM or RAD50 as well as amplification of AURKA are associated with in vitro sensitivity to PARP inhibition (11, 12, 30). Moreover, we also show, that among other genes low expression of RAD51L, BCCIP (cofactor for BRCA2), and BRCC3 as well as MSH5 or MLH1 were significantly associated with PARP inhibitor sensitivity. Although deficiency of BRCC3 or BCCIP has not been associated with PARP inhibitor response so far, it is likely to be the case due to their important function in homologous recombination (31, 32). EMSY maps to the 11q13-q14 region, is commonly amplified in ovarian cancer, and has been identified as a BRCA2-binding partner (14). It has been proposed that EMSY plays a role in homologous recombination-mediated repair of DNA DSBs; however, this has been a controversial issue. Our results confirm a significant association between amplification of EMSY and sensitivity to rucaparib and support the hypothesis that EMSY amplification may be a mechanism of BRCA2 pathway inactivation in sporadic ovarian cancers. Recent studies have shown a distinct role for PTEN in the maintenance of chromosomal integrity and repair of DNA DSBs (33). Although it has been reported that loss of PTEN sensitizes cancer cell lines to PARP inhibitors, our preclinical results do not confirm these findings (13, 34). PTEN mutations were not statistically significantly associated with in vitro sensitivity to rucaparib using the current cell line panel of 39 established human ovarian cancer cell lines. Earlier preclinical studies suggest that epigenetic silencing of the BRCA1 promoter region in the UACC3199 breast cancer cell line may be associated with sensitivity to rucaparib (18). In the current ovarian cancer cell line panel, however, we were not able to detect methylation of BRCA1/2 promoter regions. This seems to contrast clinical studies where epigenetic silencing of BRCA1 has been described to occur in approximately 10% of serous papillary ovarian cancers (10). Nevertheless, our findings do mirror earlier preclinical studies using breast cancer models, which similar to our study did not find BRCA1/2 methylation in a panel of 21 established breast cancer cell lines (35). Taken together, both of these studies suggest that BRCA1/2 promoter methylation is rarely found in established breast or ovarian cancer cell lines.

Recent clinical data indicate that patients with platinum-sensitive recurrent sporadic ovarian cancer may benefit from single-agent PARP inhibitor therapy (8, 9). In one of these studies, objective responses were seen in 10 of 20 (50%) patients with platinum-sensitive recurrent sporadic ovarian cancer, as opposed to only in 1 of 26 (4%) patients with platinum-resistant recurrent ovarian cancer (8). This association is supported by our findings, which show a statistically significant positive correlation between sensitivity to carboplatin and sensitivity to rucaparib.

It has been proposed that PARP inhibitors can potentiate the cytotoxicity of DNA-damaging agents by preventing DNA repair (36, 37). Here, we show that drug interactions with rucaparib were synergistic for topotecan, carboplatin, or doxorubicin in ovarian cancer cells both sensitive and resistant to single-agent rucaparib. As such, despite a relative insensitivity of HEY, MCAS, or OV2008 ovarian cancer cells to single-agent rucaparib, treatment with rucaparib did potentiate chemotherapy. Data from our group as well as others (36, 37) suggest that even in the presence of a functional homologous recombination pathway and a cell's ability to conduct DNA DSB repair, the homologous recombination pathway may only be able to repair part of the additional damage introduced by PARP inhibition in homologous recombination proficient cells treated with a DNA-damaging agent, but not all and thus result in enhanced cytotoxicity. Synergy was most pronounced when rucaparib was combined with topotecan, which resulted in an increase in apoptosis, DNA fragmentation, and yH2AX formation. The effect of topo-I inhibition is the induction of DNA SSBs (38), which are primarily repaired by the PARP-dependent BER pathway. In addition, it has been suggested that PARP inhibitors can enhance topo-I activity, which could sensitize cells to topoisomerase I poisons (39). The combination of topotecan with the PARP inhibitor olaparib has recently been tested in a clinical phase I study in patients with advanced solid tumors (29). This trial showed enhanced topotecan-induced myelosuppression when combined with olaparib. Of note, our data also indicate that the most pronounced synergy was seen at low topotecan drug concentrations in vitro, which may imply that efficacy can be maintained even if a dose reduction of topotecan is required (29). Taken together, the present studies suggest that rucaparib sensitivity is not restricted to ovarian cancer with BRCA1/2 mutations and, furthermore, confirms the notion that PARP inhibitors may be able to potentiate cytotoxic treatment independent of their activity as single agent. These findings support further clinical evaluation of rucaparib either as single agent or as adjunct to chemotherapy in the treatment of sporadic ovarian cancer.

S. Jones is employed as a scientist in Personal Genome Diagnostics and has ownership interest (including patents) in stock options in PGDX. V.E. Velculescu is on the board of directors and is CSO of Personal Genome Diagnostics, has ownership interest (including patents) in Personal Genome Diagnostics and Inostics, and is a consultant/advisory board member of Inostics. D.J. Slamon has a honoraria from speakers' bureau from Genentech, Sanofi-Aventis, and GlaxoSmithKline, has ownership interest (including patents) in Amgen, and is a consultant/advisory board member of Novartis Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M. Ihnen, F. Janicke, G. Los, D.J. Slamon, G.E. Konecny

Development of methodology: J. Qi, M. Chalukya, C. Ginther, V.E Velculescu, D.J. Slamon, G.E. Konecny

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Ihnen, T. Kolarova, J. Qi, K. Manivong, L. Anderson, C. Ginther, A. Meuter, B. Winterhoff, S. Jones, V.E Velculescu, S. Dandekar, D.J. Slamon, G.E. Konecny

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Ihnen, C. zu Eulenburg, T. Kolarova, J. Qi, K. Manivong, M. Chalukya, J. Dering, L. Anderson, S. Jones, N. Udar, D.J. Slamon, G.E. Konecny

Writing, review, and/or revision of the manuscript: M. Ihnen, C. zu Eulenburg, T. Kolarova, B. Winterhoff, S. Dandekar, N. Udar, F. Janicke, G. Los, D.J. Slamon, G.E. Konecny

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Ihnen, J. Qi, M. Chalukya, S. Jones, N. Venkatesan, H.-M. Rong, D.J. Slamon, G.E. Konecny

Study supervision: M. Ihnen, G.E. Konecny

G.E. Konecny has been supported in part by the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, the Thelma L. Culverson Endowed Cancer Research Fund, and the Stranahan Foundation for Translational Cancer Research and Advanced Clinical Cancer Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.