Research ArticleCancer

Androgen receptor inhibitor–induced “BRCAness” and PARP inhibition are synthetically lethal for castration-resistant prostate cancer

See allHide authors and affiliations

Science Signaling  23 May 2017:
Vol. 10, Issue 480, eaam7479
DOI: 10.1126/scisignal.aam7479

Engineering BRCAness and chemotherapeutic sensitivity

BRCA mutations impair a double-strand break DNA repair pathway that forces cells to use a PARP-dependent repair pathway. PARP inhibitors are selectively toxic to breast cancers with BRCA mutations, spurring the search for other tumors or ways in which to apply such exquisitely tumor-targeted therapy. Few other tumors have BRCA mutations as commonly as do breast tumors. However, Li et al. found that a common therapy for prostate cancer patients created a BRCA-deficient state that sensitized tumor cells to PARP inhibitors and leveraged this finding into a potential treatment strategy. Noting that the androgen receptor inhibitor enzalutamide decreased the expression of BRCA1 in prostate cancer cells, the authors treated a mouse model of prostate cancer first with enzalutamide and then with the PARP inhibitor olaparib. Sequential treatment of enzalutamide and olaparib suppressed tumor growth in these mice better than either drug by itself or when both drugs were administered at the same time. The results suggest that “BRCAness” could be therapeutically induced to provide more treatment options not only for prostate cancer patients but also for patients with other types of cancers lacking BRCA mutations.

Abstract

Cancers with loss-of-function mutations in BRCA1 or BRCA2 are deficient in the DNA damage repair pathway called homologous recombination (HR), rendering these cancers exquisitely vulnerable to poly(ADP-ribose) polymerase (PARP) inhibitors. This functional state and therapeutic sensitivity is referred to as “BRCAness” and is most commonly associated with some breast cancer types. Pharmaceutical induction of BRCAness could expand the use of PARP inhibitors to other tumor types. For example, BRCA mutations are present in only ~20% of prostate cancer patients. We found that castration-resistant prostate cancer (CRPC) cells showed increased expression of a set of HR-associated genes, including BRCA1, RAD54L, and RMI2. Although androgen-targeted therapy is typically not effective in CRPC patients, the androgen receptor inhibitor enzalutamide suppressed the expression of those HR genes in CRPC cells, thus creating HR deficiency and BRCAness. A “lead-in” treatment strategy, in which enzalutamide was followed by the PARP inhibitor olaparib, promoted DNA damage–induced cell death and inhibited clonal proliferation of prostate cancer cells in culture and suppressed the growth of prostate cancer xenografts in mice. Thus, antiandrogen and PARP inhibitor combination therapy may be effective for CRPC patients and suggests that pharmaceutically inducing BRCAness may expand the clinical use of PARP inhibitors.

INTRODUCTION

Metastatic prostate cancer (PCa) is incurable and the second leading cause of cancer deaths among men in the Western world (1). Although patients initially respond to androgen deprivation therapy, most eventually develop castration-resistant prostate cancer (CRPC) and metastasis to bone, the predominant site of advanced lethal PCa (2, 3). Abiraterone acetate and enzalutamide (ENZ), novel androgen receptor (AR) signaling inhibitors (4, 5), each extends the life of metastatic CRPC patients for 4 to 5 months, but neither of these agents is curative. Therefore, more effective therapies are needed, including novel therapeutic combinations and innovative therapy sequencing approaches.

The development of CRPC is accompanied by the accumulation of gene mutations, chromosomal translocations, and enhanced DNA repair activities that enable CRPC cells to survive and proliferate (68). Moreover, increased expression of certain DNA repair genes reflects reduced DNA repair proficiency, and these activities are associated with greater mutagenesis, adverse clinical features, and inferior patient survival rates (9). Poly(adenosine diphosphate–ribose) polymerase (PARP) inhibitors, which can abrogate base excision repair or trap PARP on the DNA, are promising therapeutic agents that show synthetic lethality against many types of cancer with BRCA1 or BRCA2 deficiencies (1013). Although a favorable response of cancers to PARP inhibition due to BRCA1 or BRCA2 mutations has been demonstrated, PARP inhibitors have not been widely used in cancers with deficiencies of other homologous recombination (HR) genes. Promising clinical studies using PARP inhibitors are under way (1416), and it is now essential to further our understanding of the HR pathway to maximize the benefit of PARP inhibitor therapy for CRPC. BRCA germline/somatic mutations/deletions account for only 12 to 20% of CRPC patients (6, 14, 17, 18); therefore, pharmaceutically induced HR deficiency becomes an attractive and clinically viable strategy that potentially expands the benefit of PARP inhibitor therapy to CRPC patients who do not have BRCA mutations.

Accumulating evidence revealed the convergence of DNA damage response (DDR) signaling pathways and AR signaling pathways in PCa (8, 1922). We previously found that an AR- and c-Myb–coregulated DDR gene signature is correlated with PCa metastasis, castration resistance, tumor recurrence, and reduced survival in PCa patients (8). The presence of a relatively large number of HR genes in this AR- and c-Myb–coregulated DDR gene signature, together with the identification of a CRPC–up-regulated HR gene signature (Fig. 1), led us to consider that pharmaceutically induced down-regulation of AR-regulated HR genes may induce HR deficiency and sensitize CRPC to PARP inhibition. Therefore, targeting AR by ENZ and PARP by olaparib (OLA) would generate synthetic lethality in CRPC. Here, we report that ENZ down-regulated a specific set of HR genes, including BRCA1, RAD51AP1, RAD51C, RAD54L, and RMI2, and this pharmaceutically induced HR deficiency synergizes with the effects of OLA to induce DNA damage–related cell death, inhibit PCa cell clonal growth, and suppress xenograft tumor growth.

Fig. 1 HR gene expression in human normal prostate and primary PCa and CRPC tissue samples.

(A and B) Box plots showing 10 CRPC–up-regulated (yellow) and 1 CRPC–down-regulated (RPA1; blue) HR genes common to two published CRPC data sets. (C and D) Heat maps showing expression of 10 commonly up-regulated HR genes in two published CRPC data sets.

RESULTS

Expression of HR-associated genes is up-regulated in CRPC

To extend our previous study showing the correlation of advanced PCa and AR- and c-Myb–coregulated DDR gene signature, in which HR-associated genes are highly represented (8), we identified specific HR genes that were up-regulated in CRPC using two public patient data sets (6, 17) and a Gene Ontology–defined set of 37 HR-associated genes. Analysis of these HR genes showed significant differential expression between hormone-naïve primary PCa and CRPC samples. Statistically altered HR genes were identified after Benjamini-Hochberg correction for multiple hypothesis testing (P < 0.05), and those common to both data sets were identified (Fig. 1, A and B). There were 10 up-regulated HR genes common to both data sets, including CHEK1, BRCA1, EXO1, BLM, RMI1, RAD54L, RAD51, LIG1, XRCC3, and RMI2, and 1 down-regulated HR gene (RPA1). Analysis of gene expression in CRPC tissue samples revealed that a set of HR genes is overexpressed in CRPC and positively correlated with CRPC development and adverse clinical outcome (Fig. 1, C and D).

ENZ, OLA, and their combination inhibit HR-associated gene expression in AR-positive PCa cells

To test our hypothesis that pharmaceutically induced down-regulation of AR-regulated HR genes may induce BRCAness and sensitize CRPC to PARP inhibition, we examined the effect of ENZ, OLA, and the combination (ENZ + OLA) on HR gene expression in multiple PCa cell lines with variable androgen dependence. Microarray analysis revealed that expressions of 15 of 37 HR genes were statistically altered in response to drug treatment (after Benjamini-Hochberg correction for multiple hypothesis testing, P < 0.05) (Fig. 2A). Among these 15 HR genes, we selected 5 genes, namely, BRCA1, RAD51AP1, RAD51C, RAD54L, and RMI2 for further investigation. Our selections were made on the basis of the following: BRCA1, RAD54L, and RMI2 are among the 10 up-regulated HR genes in CRPC in both of the independent clinical data sets (Fig. 1); these 5 genes are among most significantly altered ENZ-targeted, down-regulated HR genes; and these 5 genes are involved in different critical steps in HR-mediated DNA repair (2326). Quantitative reverse transcription polymerase chain reaction (qRT-PCR) (Fig. 2, B to D) and Western blotting (Fig. 2E) analyses showed that ENZ treatment inhibited the expression of all five genes in VCaP cells (AR-positive, the most androgen-dependent of the three cell lines used, and derived from a vertebral metastatic lesion of a patient with hormone refractory PCa) and all five genes to a lesser extent in LNCaP cells (AR-positive but less androgen-dependent than VCaP and derived from a patient with a supraclavicular lymph node metastatic lesion; this cell line also contains a point mutation in the ligand-binding domain of AR, leading to altered hormone specificity) but did not inhibit any of these genes in CWR22Rv1 cells [AR-positive and androgen-independent and derived from a primary prostate tumor that expresses mutant (H874Y) ARs and secretes low amounts of prostate-specific antigen]. Unexpectedly, OLA treatment suppressed the expression of most of these HR genes in LNCaP cells and all of these HR genes in CWR22Rv1 cells but had very little effect on these HR genes in VCaP cells. Further down-regulation of these HR genes by the combination of ENZ and OLA compared to single agent alone was evident in the LNCaP model (Fig. 2, A to E).

Fig. 2 Cell context–dependent ENZ and OLA down-regulation of HR genes in PCa cells with different androgen dependency.

(A) Heat map showing the expression of 15 HR-associated genes that were significantly altered by ENZ in AR-positive and androgen-dependent VCaP and LNCaP cells and in AR-positive and androgen-independent CWR22Rv1 cells. Significance (P < 0.05) was determined by Benjamini-Hochberg correction for multiple hypothesis testing. (B to D) qRT-PCR analysis of mRNA expression of five select HR genes in VCaP (B), LNCaP (C), and CWR22Rv1 (D) cells. Data are means ± SD from three experiments. *P < 0.05, **P < 0.01 (Wilcoxon–Mann-Whitney test). (E) Western blotting analysis of five selected HR proteins in the indicated cell lines. E, ENZ; O, OLA. Blots are representative of three or more experiments.

HR genes cooperatively contribute to DNA repair and PCa cell survival

To determine the role of these ENZ–down-regulated HR genes with regard to OLA sensitivity of PCa cells, we performed cell cycle analysis to quantitate sub-G1 (apoptotic) cells resulting from HR gene silencing and OLA combination treatment. We screened multiple small interfering RNAs (siRNAs) for each HR gene in VCaP cells, selected the two most effective siRNAs for each HR gene to analyze the biological activities in VCaP cells, and then chose the more effective one to use for the analysis in LNCaP and CWR22Rv1 cells. Our results demonstrate that each HR gene differentially contributes to resistance to OLA treatment and PCa cell survival as indicated by the increase in the sub-G1 cell population in response to knockdown of a specific HR gene (Fig. 3, A to F). Synergy analyses using two-way analysis of variance (ANOVA) (27) and Bliss independence model (28) indicated that knockdown of BRCA1, RAD51C, or RMI2 synergized with OLA in all three cell lines; knockdown of RAD51AP1 synergized with OLA in VCaP and CWR22Rv1 cells; and knockdown of RAD54L synergized with OLA in LNCaP and CWR22Rv1 cells (Fig. 3, D to F, and table S1).

Fig. 3 HR gene silencing synergizes with OLA to increase cytotoxicity to PCa cells.

(A to C) Western blotting analysis for the expression of HR genes in VCaP (A), LNCaP (B), and CWR22Rv1 (C) cells transfected with specific HR gene siRNAs (20 nM). Two siRNAs per HR gene were tested in VCaP cells, as indicated; one each was used in LNCaP and CWR22Rv1. Blots are representative of three or more experiments. (D to F) Flow cytometry analysis of sub-G1 cells after siRNA and OLA treatment in VCaP (D), LNCaP (E), and CWR22Rv1 (F) cells transfected as in (A) 24 hours before treatment with dimethyl sulfoxide (DMSO) or OLA (10 μM for 48 hours). (G) Western blotting analysis for BRCA1, RAD51C, and RMI2 proteins, γH2AX abundance, and cleaved PARP abundance by RNA interference and OLA, as indicated, in LNCaP cells. Blots are representative of three or more experiments. (H) Flow cytometry analysis of sub-G1 cells for synergistic effects of combinations of three selected HR gene siRNAs in the absence and presence of OLA in LNCaP cells. Data are means ± SD from three experiments. (I) Colony assay for synergistic effects of combinations of three selected HR gene siRNAs in the absence and presence of OLA in LNCaP cells. Data are means ± SD from three experiments. #Synergy by two-way ANOVA; ^synergy by Bliss independence analysis (table S1).

HR is a multistep biological process (2326). We hypothesized that ENZ– and/or OLA–down-regulated HR genes that function at different HR steps cooperatively contribute to DNA repair and PCa cell survival. Therefore, simultaneous knockdown of multiple HR genes should cooperatively enhance the HR deficiency phenotype and synergize the HR defect with OLA. We selected BRCA1, RAD51C, and RMI2 to test cooperative functions because these three HR genes are involved in the recruitment of early HR factors to double-strand breaks, formation of the RAD51 nucleoprotein filaments, and resolution of double Holliday junctions, respectively (2326) and because the siRNAs for each of these three HR genes showed synergy with OLA treatment in all three cell lines tested (Fig. 3, D to F). Simultaneous knockdown of two or three HR-associated genes markedly increased the abundance of γH2AX protein (a marker of DNA damage) and cleaved PARP (a marker of apoptosis) (Fig. 3G), significantly increased proportion of sub-G1 cells (Fig. 3H), and significantly reduced capacity for colony formation (Fig. 3I) compared with knockdown of each HR gene singly. Synergistic effects were observed in most of the HR-associated siRNA combinations and HR-associated siRNA and OLA combinations (Fig. 3, H and I, and table S1). Dual knockdown of BRCA1 and RMI2 showed the greatest effects (Fig. 3, G to I) among the tested dual siRNA combinations. Together, these results suggest that ENZ-regulated HR genes cooperatively contribute to DNA repair and PCa cell survival.

ENZ and/or OLA inhibits HR efficiency

γH2AX and RAD51 are two markers of DNA damage, DNA repair, and treatment sensitivity in preclinical and clinical samples (29). To determine whether ENZ and OLA treatment leads to reduced HR efficiency, we conducted double immunofluorescence staining to assess the formation of γH2AX and RAD51 foci and their colocalization in OLA + ENZ–treated LNCaP cells (Fig. 4A). Quantitative analyses demonstrated that the percentages of cells that contain γH2AX foci were significantly higher in ENZ + OLA–treated cells compared to those in OLA-treated cells (Fig. 4B, left), and the percentage of cells with both γH2AX and RAD51 foci was significantly lower in response to ENZ + OLA treatment than to OLA treatment alone (Fig. 4B, right). Foci ratio of RAD51 to γH2AX was significantly lower in ENZ + OLA–treated cells than in those treated with OLA alone (Fig. 4C). Analysis of drug-treated LNCaP cells using a quantitative PCR (qPCR)–based HR assay showed that ENZ and OLA each reduced HR efficiency and that their combination further reduced HR efficiency (Fig. 4D). Western blotting analysis of drug-treated LNCaP cells indicated that ENZ + OLA combination treatment markedly increased DNA damage or its persistence (γH2AX) and DNA damage–induced apoptosis (cleaved PARP) (Fig. 4E). Together, these data suggest that the cytotoxic effects of the ENZ + OLA combination are mediated by deficient DNA repair.

Fig. 4 ENZ and OLA suppress DNA repair in PCa.

(A) Double immunofluorescence to detect γH2AX (red) and RAD51 (green) foci and their colocalization in OLA + ENZ–treated LNCaP cells. (B and C) Quantitation (percentage) of γH2AX or RAD51 foci-positive cells and γH2AX and RAD51 foci-double positive cells (B) and the ratio of RAD51 to γH2AX foci in LNCaP cells. (D) Relative HR activity was calculated by a two-step normalization: The number of PCR cycles in assay primer group was normalized by that in universal primer group (internal control); the number of PCR cycles in treatment groups was normalized by that of control, with control = 1. Data are means ± SD from three experiments. P values were derived from Wilcoxon–Mann-Whitney Test. (E) Western blotting analysis of γH2AX and cleaved PARP in lead-in ENZ + OLA–treated cells compared to those in OLA-treated cells. Blots are representative of three experiments.

Lead-in ENZ + OLA strategy synergistically increases apoptosis and inhibits clonogenic growth in PCa cells

To optimize treatment conditions of ENZ and OLA combination, we compared a lead-in ENZ + OLA treatment strategy with a traditional concomitant ENZ + OLA strategy. In the lead-in ENZ + OLA treatment, we pretreated PCa cells with ENZ for 24 hours (lead-in time) and then with ENZ + OLA for 48 hours. This protocol achieved ENZ-mediated down-regulation of HR genes before OLA was applied. In the concomitant ENZ + OLA treatment, cells were treated with DMSO (vehicle control for ENZ) for 24 hours and then with ENZ + OLA for 48 hours. The resulting data showed that the lead-in ENZ + OLA treatment increased sub-G1 cells and inhibited colony growth more effectively than traditional concomitant ENZ + OLA treatment in VCaP and LNCaP cells (AR-positive and androgen-dependent), but not in CWR22Rv1 cells (AR-positive and androgen-independent) (Fig. 5, A to D, and table S1).

Fig. 5 A superior lead-in effect and its underlying mechanisms.

(A to C) Cell cycle analysis. Cells were pretreated with ENZ (0.5 or 1 μM) or DMSO (vehicle control) for 24 hours, followed by the treatment with DMSO, ENZ (0.5 or 1 μM) or OLA (5 or10 μM), or ENZ + OLA for 48 hours. (D) Colony assay. Optimal concentrations of ENZ and OLA were determined for each cell line by drug titration (fig. S2). Cells were pretreated with ENZ (VCaP, 100 nM; LNCaP, 150 nM; CWR22Rv, 1 μM) or DMSO (vehicle) for 24 hours, followed by treatment with ENZ, OLA (VCaP, 2.5 μM; LNCaP, 1.5 μM; CWR22Rv1, 500 nM), ENZ + OLA, or DMSO for 10 to 20 days depending on the cell line. #Synergy by two-way ANOVA; ^synergy by Bliss independence model. a.u., arbitrary units. (E) Functional illustration of two ENZ–down-regulated and OLA–up-regulated antiapoptotic genes SGK1 and TNFAIP8. (F) Western blotting analysis of selected pro- and antiapoptotic proteins from LNCaP cells treated with DMSO, ENZ, OLA, E + O_C (concomitant), or E + O_L (lead-in). (G) Western blotting analysis for SGK1 and TNFAIP8 in control and siRNA-transfected LNCaP cells. Blots are representative of three experiments. (H) Sub-G1 cell analysis in control, SGK1 siRNA-, or TNFAIP8 siRNA-transfected in LNCaP cells. Data are means ± SD from three experiments. *P < 0.05 by Wilcoxon–Mann-Whitney test [(A to C) versus the corresponding concomitant condition and (H) versus the control siRNA under DMSO condition].

Because the lead-in ENZ + OLA treatment protocol further increased the number of apoptotic cells compared to the concomitant ENZ + OLA protocol, we considered differential ENZ versus OLA apoptotic activities combined with increased exposure to ENZ during the lead-in treatment phase as underlying cooperative mechanisms for the lead-in effect. Using unambiguous pro- and antiapoptotic gene subsets derived from the Gene Ontology–defined apoptosis gene set (GO: 0006915), we pooled microarray expression data of these pro- and antiapoptotic genes, adjusted P value using Benjamini-Hochberg procedure for multiple hypothesis testing, and identified statistically significant (P < 0.05) alterations in gene expression in ENZ, OLA, or lead-in ENZ + OLA–treated samples compared to those treated with vehicle (tables S2 and S3). In addition to overlapping subsets of ENZ- and OLA-induced up-regulated proapoptotic genes and overlapping subsets of ENZ- and OLA-induced down-regulated antiapoptotic genes, which could contribute to synergistic effects from ENZ and OLA, we found that two of the antiapoptotic genes, encoding serum/glucocorticoid-regulated kinase 1 (SGK1) and tumor necrosis factor–α (TNF-α)–induced protein 8 (TNFAIP8), were down-regulated by ENZ but up-regulated by OLA, which led us to speculate that the lead-in protocol would enable maximal ENZ-induced suppression and delay OLA up-regulation of these specific antiapoptotic genes. Both SGK1 and TNFAIP8 reportedly participate in apoptosis evasion, cell survival, chemoresistance, and tumor progression (Fig. 5E) (3034). Western blotting analysis validated select ENZ– and OLA–up-regulated proapoptotic genes and ENZ–down-regulated and OLA–up-regulated SGK1 and TNFAIP8 (Fig. 5F). To validate the pro–cell survival roles of SGK1 and TNFAIP8, we knocked down each in LNCaP cells (Fig. 5G) and performed cell cycle analysis to assess sub-G1 cell distribution. SGK1 knockdown alone caused only a slight increase in the proportion of sub-G1 cells but significantly increased the proportion of sub-G1 cells when combined with OLA, whereas TNFAIP8 knockdown alone significantly increased the proportion of sub-G1 cells that was further increased by combination with OLA (Fig. 5H). Our data suggest that, in addition to down-regulation of specific HR genes, down-regulation of SGK1 and TNFAIP8 expression by ENZ before OLA treatment, in part, contributes to the superior therapeutic effect of the lead-in ENZ + OLA strategy.

Lead-in ENZ + OLA treatment strategy synergistically suppresses PCa growth

To test our finding of synergism in the lead-in ENZ + OLA treatment strategy in vivo, we analyzed ENZ or OLA alone and concomitant or lead-in ENZ + OLA combination treatment using a subcutaneous patient-derived xenograft (PDX) tumor model [MDA PCa 133-4, AR-positive, and implanted in precastrated severe combined immunodeficiency (SCID) mice] and two orthotopic PCa cell line xenograft models (VCaP, AR-positive, and androgen-dependent; CWR22Rv1, AR-positive, and androgen-independent) (Fig. 6A). In the MDA 133-4 subcutaneous model, treatment with ENZ or OLA alone was sufficient to suppress tumor growth, indicated by reduced tumor volumes (Fig. 6B) and tumor weights (Fig. 6C), with ENZ demonstrating superior efficacy among the two single agents. The in vivo lead-in treatment with ENZ + OLA, which mimicked that used in cultured cells, achieved significantly greater tumor suppression than treatment with either OLA or ENZ alone or with concomitant ENZ + OLA treatment (Fig. 6, B and C). Synergy analysis revealed a synergistic therapeutic effect in the lead-in ENZ + OLA group (Fig. 6, B and C, and table S1). Analysis of xenograft tumor samples demonstrated significantly reduced BRCA1 protein abundance and Ki67 staining in tumors from mice treated with the lead-in ENZ + OLA protocol compared to those from mice treated with vehicle, ENZ, or OLA alone, or concomitant ENZ + OLA (Fig. 6, D and E). Similar results were observed using VCaP orthotopic xenograft model (Fig. 7). ENZ had superior efficacy in the MDA 133-4 and VCaP models; by contrast, in the CWR22Rv1 model, OLA showed superior efficacy, and the lead-in ENZ + OLA and concomitant ENZ + OLA showed efficacy similar to that of OLA alone (Fig. 8, A to C). Although BRCA1 abundance was relatively unchanged (Fig. 8D), Ki67 labeling was significantly reduced in CWR22Rv1 xenograft samples from OLA- or lead-in ENZ + OLA–treated mice (Fig. 8E).

Fig. 6 Scheme of in vivo experiments and results of MDA PCa 133-4 subcutaneous model.

(A) Scheme of in vivo experiments for VCaP and CWR22Rv1 orthotopic models and MDA 133-4 subcutaneous model. (B and C) Suppression of MDA PCa 133-4 subcutaneous tumor growth by ENZ and OLA. Tumor growth (B) assessed 1 day before treatment and through 27 days of therapy and tumor weight (C) assessed at day 27 of treatment in MDA 133-4 subcutaneous xenografts. Data are means ± SE from 10 or more mice in each group. P values were derived from ANOVA t test. #Synergy determined by two-way ANOVA; ^synergy determined by Bliss independence analysis. VH, vehicle. (D and E) Immunostaining for BRCA1 (D) and Ki67 (E) at day 14 of the indicated treatment. Data are means ± SE from seven or more mice; P values were derived from Wilcoxon–Mann-Whitney test.

Fig. 7 Suppression of VCaP orthotopic xenograft tumor growth by ENZ and OLA treatment.

(A to C) Representative luminescent images (A), tumor growth (B), and tumor weight (C) of VCaP orthotopic xenografts treated as indicated. Images are representative of experiments, and data are means ± SE from 17 or more mice in each group. P values were derived from ANOVA t test. ^Synergy determined by Bliss independence analysis. (D and E) Immunohistochemcal staining for BRCA1 (D) and Ki67 (E) at 7 (BRCA1) or 10 (Ki67) days of the indicated treatment. Data are means ± SD from four or more mice. P values were derived from Wilcoxon–Mann-Whitney test.

Fig. 8 Suppression of CWR22Rv1 orthotopic xenograft tumor growth by ENZ and OLA treatment.

(A) Representative bioluminescent images from 10 or more mice in each group. (B and C) Tumor growth (B) and weight (C) of CWR22Rv1 orthotopic xenografts from mice treated as indicated. Data are means ± SE. ^Synergy by Bliss independence analysis. (D and E) Immunohistochemical staining for BRCA1 (D) and Ki67 (E) in tumors removed from mice after 8 days of the indicated treatment. Data are means ± SE from six or more mice. P values were derived from Wilcoxon–Mann-Whitney test.

DISCUSSION

The utilization of PARP inhibitors such as OLA has rapidly evolved from in vitro studies to clinical applications (16, 35, 36). However, only recently has limited information become available for PCa. Recent reports include the results from preclinical models (8, 37), a phase 1 clinical trial using OLA as single agent (15), a phase 1/2 clinical trial using veliparib in combination with temozolomide (38, 39), and a more recent phase 2 trial using OLA (14). It is noteworthy that Mateo et al. extended the BRCAness concept to mutations/deletions of genes involved in HR regulation (ATM), HR factor nuclear localization (PALB2), and chromatin remodeling during HR DNA repair (HDAC2), which conceptually supports our finding that BRCA1/2 and other HR genes involved in different stages of HR DNA repair cooperatively contribute to OLA sensitivity and our proposal that pharmaceutically induced HR deficiency as a therapy concept for future clinical applications of OLA. By demonstrating that 10 specific HR genes are up-regulated in CRPC in two public data sets and that these up-regulated HR genes represent therapy targets for ENZ + OLA combination treatment, we introduce a specific biomarker panel for the use of PARP inhibitors and expand the concept of BRCAness to the concept of HR deficiency for CRPC as regard to clinical applications of PARP inhibitors for synthetic lethal therapy.

HR occurs through a series of steps (23, 25, 26), and the five ENZ–down-regulated HR genes that we identified are involved in almost all the major steps. Our data suggest that each ENZ–down-regulated HR gene has a role in HR and the maintenance of PCa cell survival but that BRCA1, RAD51C, and RMI2, in particular, cooperatively contribute to DNA repair and PCa cell survival, perhaps through their roles during sequential steps in the HR pathway. The superior performance of dual and triple knockdown underscores redundancy among HR mediators that may oppose PARP inhibitor efficacy. Therefore, ENZ-induced suppression of multiple HR genes is a viable strategy to improve the efficacy of PARP inhibitors. The concept, if applied clinically, may extend the benefit of PARP inhibition therapy to CRPC patients who do not bear germline or somatic mutations in or deletion of BRCA1/2 or other HR genes.

We also demonstrated that lead-in ENZ + OLA has a clear advantage compared to concomitant ENZ + OLA. In addition to down-regulation of a specific set of HR genes by ENZ before OLA treatment, this strategy also allows maximal ENZ suppression of and delay of OLA up-regulation of specific antiapoptotic genes. SGK1 and TNFAIP8 are two ENZ–down-regulated and OLA–up-regulated antiapoptotic genes identified in this study, and knockdown of either gene was proved to have significant synergistic effect with PARP inhibition as regard to apoptotic cell death. These results suggest that down-regulation of SGK1 and TNFAIP8 by ENZ before OLA treatment is, at least, one of the mechanisms underlying the superior effect of lead-in ENZ + OLA over concomitant ENZ + OLA; they are also consistent with growing evidence that links SGK1 and TNFAIP8 to cancer cell survival, metastasis, and chemo- and radioresistance (3034).

The results of our xenograft studies provided particularly compelling support for the clinical utility of the ENZ + OLA combination treatment approach for advanced PCa or CRPC. Synergistic growth suppression was associated with ENZ-sensitive models (that is, MDA PCa133-4 and VCaP). The lead-in approach demonstrated clear superiority of the ENZ + OLA combination treatments for the ENZ-sensitive models compared to the traditional concomitant application protocols. Suppression of nuclear BRCA1 levels by the ENZ + OLA combination was associated with treatment efficacy in these models. Overall, these results have direct relevance for future ENZ + OLA clinical trials, which should be considered in light of our results.

Our data also showed that OLA strongly suppressed HR genes in androgen-independent CWR22Rv1 cells, indicating a cell context–dependent role for OLA in HR gene regulation, and it is noteworthy that our results suggest a transition from AR- to PARP-dependent regulation of HR genes. Future studies are warranted to address these possibilities. Although ENZ failed to synergize with OLA in this cell line model, OLA can down-regulate HR genes in this type of PCa cells and compensate for the loss of ENZ effect.

Overall, our study provides new insights into the molecular functions of ENZ, adds mechanistic rationale to the ENZ + OLA combination therapy, and underscores the importance of an expanded BRCAness concept in future clinical usage of OLA. Our findings support a randomized clinical trial for OLA as single agent or in combination with other agents in unselected patients with CRPC. Future investigations using baseline and OLA-treated patient tissue samples are warranted to link the genomic and transcriptomic landscape, including these ENZ–down-regulated HR genes, with patient response to OLA therapy.

MATERIALS AND METHODS

Sources of cell lines and PDXs

VCaP and LNCaP (American Type Culture Collection) and CWR22Rv1 (from F. M. Sirotnak of Memorial Sloan Kettering Cancer Center) cell lines were validated by short tandem repeat DNA fingerprinting with the AmpFlSTR Identifiler PCR Amplification Kit (Applied Biosystems) in MD Anderson’s Characterized Cell Line Core Facility. Genetic background (genomic alterations and copy number variations) of these cell lines, VCaP-luc, CWR22Rv1-luc, and MDA 133-4 PDX model was analyzed using MDA T200 platform and summarized in fig. S1.

Gene expression microarray

Complementary DNA (cDNA) microarray was performed using Illumina TotalPrep RNA Amplification Kit (Thermo Scientific) and Illumina Human HT-12 v4.0 Expression BeadChip Kit (Illumina). Microarray data were deposited in the Gene Expression Omnibus (GEO) database (accession number GSE69249).

Bioinformatics and biostatistics analyses

Samples of Grasso et al. [GSE35988; (6)] were profiled on two different Agilent Whole Genome Microarrays. The two expression matrices (GPL6480 and GPL6848) were combined for bioinformatics/biostatistics analyses because no obvious batch effect was detected by principal component analysis. Samples of Taylor et al. [GSE21034; (17)] were profiled on Affymetrix Human Exon 1.0 ST arrays, and transcript-level expression matrix (GPL10264) was used for the analyses. The transcript accession numbers were converted to gene symbols using University of California, Santa Cruz reference gene annotation file, and median expression level was used for genes with multiple transcripts. To compare the expression of 37 HR genes between normal prostate and hormone-naïve PCa and CRPC samples, a linear model was fit to each of the genes, followed by Tukey’s all-pair comparisons of the independent variable using R package “multcomp.” The overall P values were adjusted by Benjamini-Hochberg procedure. Genes with significant overall P value (<0.05) and also with significant P value when comparing CRPC either to hormone-naïve PCa or to normal prostate samples were selected from both Grasso and Taylor data sets. The commonly significantly up-regulated and down-regulated genes were used for box plots (Fig. 1, A and B). The expression profile of the 10 commonly up-regulated genes in the samples of Grasso et al. (Fig. 1C) and Taylor et al. (Fig. 1D) was shown in semisupervised hierarchical clustering heat maps. A similar statistical analysis was performed in cell line microarray data. The mean expression level was used for genes with multiple probes. The expression profile of genes with significant overall P value and also significant P value when comparing ENZ + OLA to DMSO for any of the three cell lines was shown in supervised hierarchical clustering heat maps (Fig. 2A). Genes associated with apoptosis were obtained by the Gene Ontology term “apoptosis” (GO: 0006915) and manually curated to characterize the pro- and antiapoptosis gene sets on the basis of PubMed gene descriptions. Difference of gene expression across various treatments was assessed using ANOVA in these gene sets. The P values were adjusted for multiple testing by Benjamini-Hochberg (false discovery rate) method. Genes with adjusted P values less than 0.05 are defined as significant. Further, Tukey’s test was used to perform multiple comparison assessment with all the contrasts of interest in the data set.

Quantitative reverse transcription polymerase chain reaction

Total RNA from cancer cell lines was extracted with the RiboPure RNA extraction kit (Thermo Fisher Scientific). Reverse transcription reactions were carried out with the High-Capacity cDNA Archive Kit (Applied Biosystems) according to the manufacturer’s protocol. Real-time PCR was performed using the ABI PRISM 7000 sequence detection system (Applied Biosystems) according to the manufacturer’s instructions. The relative quantity of mRNA was determined by the ΔΔCT (comparative change in cycle threshold) method as described by the manufacturer and normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) RNA in the same cDNA preparation. Primers and probes used for qRT-PCR were purchased from Integrated DNA Technologies and are listed in table S4.

Immunohistochemical analysis

Antibodies to BRCA1 (Abcam, M110) and Ki67 (Santa Cruz Biotechnology, SC15401) were used for immunostaining on formalin-fixed paraffin-embedded tissue slides from the mouse PCa xenografts. Briefly, rehydrated slides were microwave-heated for 20 min in citrate buffer (10 mM, pH 6.0) for antigen retrieval. They were then incubated with 1% H2O2 for 10 min to inactivate endogenous horseradish peroxidase (HRP). After blocking with serum-free protein block (Dako, X0909), they were incubated with the primary antibodies (BRCA1, 1:100; Ki67, 1:100) for 90 min at room temperature, followed by incubation with HRP-conjugated secondary antibody (Dako, K4061) for 40 min at room temperature. The immunoreaction products were visualized with 3,3′-diaminobenzidine/H2O2 solution. For quantitative analyses, 15 to 20 microscopic fields (at 200×) from each specimen were randomly selected, and immunostaining results were evaluated by investigators (G.Y. and D.K.) who were blinded to the treatment information of a specimen, and the nuclear BRCA1 or Ki67 labeling rates were recorded.

Lead-in ENZ + OLA treatment in cultured cells

VCaP, LNCaP, and CWR22Rv1 cells were pretreated with 1 μM ENZ or DMSO vehicle control for 24 hours, followed by treatment with ENZ (1 μM), OLA (10 μM for VCaP and LNCaP and 5 μM for CWR22Rv1), ENZ + OLA, or DMSO for 48 hours. The selection of OLA concentrations was based on the sensitivity of PCa cell lines to OLA.

Western blotting

Proteins were separated by electrophoresis using 4 to 20% gradient Mini-PROTEAN TGX Gels (Bio-Rad) and transferred onto a nitrocellulose blotting membrane (GE Healthcare Life Sciences). Blots with interested proteins were blocked with tris-buffered saline buffer containing 0.1% Tween 20 and 5% nonfat milk for 1 hour and incubated with specific primary antibody overnight in a cold room rocker. After incubation with HRP-conjugated secondary antibody, blots were incubated with SuperSignal West Dura Extended Duration Substrate (Thermo Scientific) for 2 min and imaged with ChemiDoc MP Imaging System (Bio-Rad). Antibodies against BRCA1 (1:2000), RAD51AP1 (1:200), RAD51C (1:200), RAD54 (1:200), RMI2 (1:2000), and TNFAIP8 (1:500) were purchased from Abcam. Antibodies against γH2AX (1:250), cleaved PARP (1:500), SGK1 (1:250), and GAPDH (1:10,000) were purchased from Cell Signaling Technology.

HR assay

HR assays were performed using LNCaP cells and an HR Assay Kit (Norgen Biotek) according to the manufacturer’s suggested protocol. Briefly, LNCaP cells were transfected with dl-1 and dl-2 vectors that carried a mutated lacZ gene in each plasmid using FuGENE Transfection Reagent (Roche). After 6 hours of plasmid transfection, cells were pretreated with ENZ or DMSO vehicle control for 18 hours and then treated with DMSO, ENZ, OLA, or ENZ + OLA for 24 hours. Total DNA was isolated, and qPCR analysis was performed using universal primers (for internal normalization) and assay-specific primers. The PCR cycle numbers for each sample using assay-specific primers was first normalized with respect to that in each sample using universal primers and then normalized with respect to control (control = 1).

RNA interference

Cells were seeded 1 day before siRNA transfection [six-well format (for RNA or protein preparation): VCaP, 1 × 106 per well; LNCaP, 5 × 105 per well; CWR22Rv1, 2 × 105 per well; 24-well format (for cell cycle analysis): one-fifth of cell numbers of six-well format). siRNA transfection was performed with 20 nM siRNA using RNAiMAX (Life Technologies) according to the manufacturer’s protocol. The sources and sequences of siRNAs are listed in table S5.

Flow cytometry analysis

Flow cytometry analysis was performed as described previously (8). Briefly, cells were seeded onto 24-well plates 1 day before the drug treatment or siRNA transfection. Flow cytometry analysis was performed 48 hours after the drug treatment or siRNA transfection. In the case of siRNA and drug combination, siRNA transfection was performed 1 day before the drug treatment, and flow cytometry analysis was performed 48 hours after the drug treatment. Cells were harvested and prepared as a single-cell suspension, washed with phosphate-buffered saline, incubated with a staining buffer containing 0.1% sodium citrate, 0.1% Triton X-100, and propidium iodine (50 mg/ml) for 30 min at 4°C, and proceeded to cell cycle analysis with a FACSCanto II flow cytometer (BD Biosciences). Cell cycle profiles and quantitative data were obtained using FlowJo software (Tree Star Inc.).

Clonogenic assay

PCa cells were seeded at a low density in six-well plates (VCaP, 1 × 105 per well; LNCaP, 2 × 104 per well; CWR22Rv1, 5 × 103 per well) and grown in normal growth medium for 10 to 20 days, and colonies were stained with 0.5% crystal violet and counted as described previously (8). Experiments were performed in triplicate, and data were means ±SD from three or more independent experiments.

Xenograft tumor assays

For the MDA 133-4 PDX model, equal-sized tumor pieces were implanted into subcutaneous pockets in the flanks of 6- to 8-week-old male CB17 SCID mice (Charles River Laboratories International Inc.). VCaP cells (3 × 106) and CWR22Rv1 cells (2 × 106) were injected orthotopically into the prostates. When tumors reached an approximate volume of 30 mm3, tumor-bearing mice were randomly placed in experimental subgroups to receive ENZ, OLA, ENZ + OLA (concomitant or lead-in: 4-day pretreatment of ENZ before combined treatment with ENZ + OLA), or a vehicle control (ENZ, 10 mg/kg every day orally; OLA, 40 mg/kg per day, 5 days each week, intraperitoneally). Subcutaneous tumor growth was monitored twice per week by caliper measurement, and orthotopic tumor growth was monitored weekly by bioluminescent images. Mice were sacrificed, and tumor wet weights were determined. A subset of mice was euthanized, tumor specimens were collected, and formalin-fixed paraffin-embedded tissue slides were prepared for immunohistochemical analysis of BRCA1 and Ki67 from tumors taken before or after 8 to 14 days of treatment.

Statistical analysis

ANOVA was used to assess the difference of gene expression, and Benjamini-Hochberg correction for multiple hypothesis testing was used to adjust the P values in the determination of statistically altered HR genes in human PCa patient samples and in drug-treated PCa cell line samples and statistically altered pro- or antiapoptotic genes in drug-treated PCa cell line samples. Wilcoxon–Mann-Whitney test was used for data with non-normal distributions or data with small sample sizes, including qRT-PCR, cell cycle analysis, colony assay, and immunohistochemical analyses. ANOVA t test was used for analysis of tumor growth and tumor wet weights. Synergistic effects were determined by two-way ANOVA tests (27) and by the Bliss independence model (28).

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/10/480/eaam7479/DC1

Fig. S1. Sequence and copy number variation analysis using T200 platform.

Fig. S2. Titration of ENZ and OLA for colony assays.

Table S1. Determination of synergy.

Table S2. Expression of proapoptotic genes.

Table S3. Expression of antiapoptotic genes.

Table S4. Primers and probes for qRT-PCR.

Table S5. siRNA sources and sequences.

REFERENCES AND NOTES

Funding: This work was supported, in part, by the National Cancer Institute grants R0150588 (to T.C.T.) and 5P50 CA140388, the Prostate Cancer Specialized Program of Research Excellence at The University of Texas MD Anderson Cancer Center, the National Cancer Institute Support Grant CA16672, and the MD Anderson Cancer Center Support Grant. G.C.M. is supported by a grant from the Michael and Susan Dell Foundation. Author contributions: T.C.T. and L.L. conceived and designed the study and wrote the paper. L.L., J.W., X.Z., W.C., and J.Y. performed the cell culture studies. S.K., S.P., J.W., Y.L., S.B., J.H.S., G.E.G., and T.K. performed the xenograft model experiments. G.Y. and D.K. performed the immunohistochemical analysis. B.M.B., G.C.M., W.W., and A.K.A. performed the gene expression analysis of CRPC patient tissue samples and integrated genomic analysis of cell lines and xenograft models. J.K., P.G.C., A.M.A., and C.J.L. provided clinical insight and concept development. P.T. performed the pathology evaluation of human PCa samples. J.-S.L. and H.-S.L. performed the microarray experiments. T.H. and C.T. provided intellectual contributions. G.E.G. revised the manuscript. All authors contributed to the data analysis. Competing interests: The authors declare that they have no competing financial interests. Data and materials availability: Microarray data are deposited in the GEO (www.ncbi.nlm.nih.gov/geo/) (accession number GSE69249).
View Abstract

Navigate This Article