Research ArticleCancer

Targeting Poly(ADP-Ribose) Polymerase and the c-Myb–Regulated DNA Damage Response Pathway in Castration-Resistant Prostate Cancer

Sci. Signal.  20 May 2014:
Vol. 7, Issue 326, pp. ra47
DOI: 10.1126/scisignal.2005070


Androgen deprivation is the standard treatment for advanced prostate cancer (PCa), but most patients ultimately develop resistance and tumor recurrence. We found that MYB is transcriptionally activated by androgen deprivation therapy or genetic silencing of the androgen receptor (AR). MYB silencing inhibited PCa growth in culture and xenografts in mice. Microarray data revealed that c-Myb and AR shared a subset of target genes that encode DNA damage response (DDR) proteins, suggesting that c-Myb may supplant AR as the dominant regulator of their common DDR target genes in AR inhibition–resistant or AR-negative PCa. Gene signatures including AR, MYB, and their common DDR-associated target genes positively correlated with metastasis, castration resistance, tumor recurrence, and decreased survival in PCa patients. In culture and in xenograft-bearing mice, a combination strategy involving the knockdown of MYB, BRCA1, or TOPBP1 or the abrogation of cell cycle checkpoint arrest with AZD7762, an inhibitor of the checkpoint kinase Chk1, increased the cytotoxicity of the poly[adenosine 5′-diphosphate (ADP)–ribose] polymerase (PARP) inhibitor olaparib in PCa cells. Our results reveal new mechanism-based therapeutic approaches for PCa by targeting PARP and the DDR pathway involving c-Myb, TopBP1, ataxia telangiectasia mutated– and Rad3-related (ATR), and Chk1.


Androgen deprivation therapy (ADT), either through drugs or orchiectomy (castration), is the standard treatment to suppress the growth of advanced prostate cancer (PCa), but most patients ultimately develop resistance and tumor growth returns. The role of the androgen receptor (AR) in the development of castration-resistant PCa (CRPC) is complex and, despite decades of research, remains poorly understood (1, 2). Increased AR expression and stimulation of specific AR target genes contribute to resistance through various mechanisms (3, 4). Castration resistance is also associated with the derepression of a specific set of AR target genes (3). A subset of aggressive tumors that display clinical features characteristic of small cell prostate carcinoma show complete loss of AR expression (5). CRPC also exhibits resistance to other therapeutic agents, including drugs that inhibit microtubules or receptor kinases (6, 7). Understanding the selection mechanisms and genetic pathways associated with drug resistance remains one of the most important problems in developing potentially curative therapies for advanced PCa.

MYB encodes c-Myb, a transcription factor with diverse cellular functions, including hematopoiesis, cell proliferation, differentiation, survival, and tumorigenesis (8). Gene fusion and copy number alterations of MYB suggest an oncogenic role in the progression of some breast, prostate, and head and neck cancers (9, 10). c-Myb increases the abundance of multiple genes associated with the progression of various malignant cells, including genes encoding proteins regulating proliferation (MYC, CCNA1, CCNB1, and CCNE1), survival (BCL2, HSPA5, HSP70, AURKA, and TPX2), and invasion and metastasis (SNAI2, MMP1, MMP9, and CSTD) (8, 1113). Although c-Myb abundance has recently been associated with PCa cell proliferation, survival, and invasion in vitro (14), the regulation, function, and mechanistic associations of c-Myb in PCa remain largely unknown.

Frequent DNA translocations and deletions arise in a highly interdependent manner in the PCa genome (15). For example, a large percentage of PCas harbor fusions of TMPRSS2 to ERG or ETV1, which are implicated in mediating advanced PCa (16). Maintenance of genomic integrity by the DNA damage response (DDR) is critical to preventing tumorigenesis, and defects in DDR-associated genes are common in multiple malignancies, including PCa (17). The loss of components of one DNA repair pathway may be compensated for by the increased activity of other pathways, which can provide necessary “stress support” for genetically unstable cancer cells (18). However, endogenous deficiencies in or targeting specific DDR-associated genes can present opportunities for cancer treatment. Poly[adenosine 5′-diphosphate (ADP)–ribose] polymerase 1 (PARP1) is a key protein in the regulation of multiple DNA repair processes (19, 20). Numerous studies have demonstrated the synthetic lethality of using PARP1 inhibitors on cancer cells that have a deficiency in the breast cancer–associated proteins 1 and 2 (BRCA1/2) (21, 22). It was also reported that pharmacological inhibition of PARP1 inhibits ETS-positive, but not ETS-negative, PCa xenograft growth (23). The cell cycle checkpoint protein Chk1 has also been shown to be a central player in the DDR network. Upon sensing DNA damage or replication stress, the ataxia telangiectasia mutated– and Rad3-related (ATR) kinase is activated, which in turn phosphorylates Chk1, leading to the activation of G2-M checkpoint and DNA repair (24, 25). Here, we tested (i) whether increased c-Myb abundance in advanced PCa is related to ADT or impairment of AR signaling; (ii) whether AR and c-Myb contribute to the DDR and correlate with PCa progression; and (iii) whether targeting specific PARP- and c-Myb–regulated DDR signaling pathways can generate synergistic cytotoxicity to PCa. Our results reveal a new mechanism-based, therapeutic strategy for PCa based on targeting PARP and the c-Myb–TopBP1–ATR–Chk1 signaling pathway.


Expression of MYB is activated by ADT or impairment of AR signaling

In an effort to identify molecular-pathologic events associated with the development of CRPC, we initially performed candidate gene expression profiling of bone metastases derived from metastatic CRPC patients and compared them to matched primary tumors or benign prostate tissue. Our analysis demonstrated significantly higher MYB expression in bone metastases, suggesting that c-Myb may play a role in the development of CRPC. By immunohistochemical (IHC) analysis, we found significantly increased c-Myb abundance in bone metastases of CRPC patients compared with their primary tumors or in benign prostate tissue (Fig. 1A). Analysis of independently published gene expression microarray data that included 4 normal bone marrow specimens, 22 local primary PCa specimens, and 29 metastatic PCa specimens (GSE32269) (26) revealed that MYB gene expression was negatively correlated with that of AR (Fig. 1B). This negative correlation was markedly increased in bone metastatic prostate tumors compared with local primary tumors (fig. S1). By IHC analysis of patient-derived xenograft (PDX) tissue microarrays, we also found increased abundance of c-Myb in AR-negative human PCa variants, including large neuroendocrine prostate carcinoma and small cell prostate carcinoma, whereas both AR-positive PDXs had lower abundance of c-Myb (Fig. 1C). Our data together with the published data suggest that androgen-AR signaling may suppress the expression of MYB.

Fig. 1 Activation of c-Myb in CRPC and under experimental conditions that impair AR signaling.

(A) IHC for c-Myb in human PCa and metastases. Numbers of specimens are given across the top. NL, normal prostate; Bone mets, bone metastasis. P value was derived using Kruskal-Wallis rank testing. Scale bars, 80 μm. (B) Correlation analysis of MYB and AR expression in a published data set (GSE32269). (C) Human PDX tissue microarray. AdCa, adenocarcinoma; LCNEC, large cell neuroendocrine carcinoma; SCPC, small cell prostate carcinoma. X axis labeling (a to i) refers to specimens listed in table S2. (D and E) MYB mRNA or c-Myb and AR protein abundance assessed by (D) qRT-PCR analysis or (E) Western blotting analysis in lysates from PCa cells that were untreated (Un), mock-transfected (Mock), or transfected with 20 nM siAR or negative control siRNA (siNC). (F and G) As in (D) and (E), in lysates from PCa cells cultured in regular medium (RM), CSS, or CSS with 10 nM R1881 (CSS + R) for 48 hours. (H) Western blotting analysis for c-Myb after treatment with 1 μM enzalutamide (ENZ) for 48 hours. (I) qRT-PCR analysis 48 hours after R1881 treatment. (J) Summary of AR regulation of c-Myb. Data in (D), (F), and (I) are means ± SD from three or more independent experiments. *P < 0.05 (Wilcoxon rank sum test). Western blots in (E), (G), and (H) are representative of three or more independent experiments. a.u., arbitrary units.

To test this concept, we examined MYB mRNA and c-Myb protein abundance in a series of experimental conditions that included androgen withdrawal and impairment of AR signaling. Analysis by gene expression microarray (fig. S2), quantitative reverse transcription polymerase chain reaction (qRT-PCR) (Fig. 1D), and Western blotting (Fig. 1E) showed that AR silencing using a targeted small interfering RNA (siRNA) induced MYB mRNA and c-Myb protein abundance in AR-positive, androgen-dependent VCaP and LNCaP, but not in AR-positive, castration-resistant CWR22Rv1 PCa cells. AR suppression of c-Myb abundance in VCaP and LNCaP cells was androgen-dependent: charcoal-stripped serum (CSS) and the second-generation AR inhibitor enzalutamide increased c-Myb mRNA and protein abundance, whereas the synthetic androgen R1881 strongly suppressed it (Fig. 1, F to H). Androgen titration experiments revealed that physiological concentrations of androgen (≥1 nM R1881), but not castration-like concentrations (≤10 pM), could efficiently suppress MYB expression (Fig. 1I), suggesting that the inhibition of AR activates c-Myb in PCa cells (Fig. 1J).

To identify the promoter region responsible for the suppression of MYB by androgen-AR signaling, we constructed a series of MYB-luciferase reporters composed of different lengths of MYB promoter DNA with or without exon 1 and/or intron 1 (Fig. 2A) and performed luciferase reporter assays in LNCaP cells treated with CSS or CSS in combination with R1881. We found that the minimum DNA fragment for the androgen-AR suppression of MYB expression is the proximal 325 nucleotides of MYB promoter (−734 to −281) (Fig. 2B). Our data also revealed androgen-independent negative regulatory elements in exon 1 and intron 1 because MYB expression was lower in all reporters with exon 1 or intron 1 than in the MYB 325-nucleotide promoter fragment (pro325nt) (Fig. 2B). To determine whether androgen-AR signaling suppressed MYB transcription through AR direct binding to or modification of the chromatin architecture of this 325-nucleotide promoter fragment, we used chromatin immunoprecipitation (ChIP) assays to assess the binding activity of AR, RNA polymerase II, and lysine-specific demethylase 1 (LSD1), which is reportedly involved in AR autosuppression (3), and the presence of transcription markers. We found increased amounts of the transcription inactivation marker trimethylated histone 3 Lys27 (H3K27), and decreased amounts of RNA polymerase II and the transcription activation markers trimethylated H3K4 and acetylated H3K27 in the 325-nucleotide MYB promoter region in response to R1881 stimulation. In contrast, the amounts of AR and LSD1 binding to this promoter region were very low and relatively unchanged by R1881 (Fig. 2C). Overall, these data indicate that AR may transcriptionally suppress MYB, in part, through chromatin modification of the MYB promoter.

Fig. 2 AR transcriptionally suppresses MYB through modification of chromatin architecture on MYB promoter.

(A) Structure of different MYB promoter luciferase reporters. (B) Luciferase reporter assays in LNCaP cells transfected with MYB promoter reporters or the control pGL3-Basic empty vector and β-gal–expressing vector (internal control) for 24 hours, followed by treatment with CSS or CSS and 10 nM R1881 (CSS + R) for 24 hours. RLU, relative luminescence units. (C) ChIP assays on the MYB promoter in LNCaP cells treated with or without 10 nM R1881 in CSS medium for 24 hours. Data are means ± SD of percentage of input from three independent experiments. *P < 0.05, Wilcoxon rank sum test. Pol II, RNA polymerase II; M3H3K27, trimethylated H3K27; M3H3K4, trimethylated H3K4; AceH3, acetylated H3K27.

MYB gene silencing significantly suppresses PCa growth in vitro and in vivo

To gain insight into the functions of c-Myb in PCa progression, we tested the effects of c-Myb knockdown on cell survival, cell motility, and clonogenic growth in cultured PCa cells using siRNA. Suppression of MYB (Fig. 3A) significantly increased the proportion of sub-G1 cells (Fig. 3B) and significantly reduced viability assessed as relative cell proliferation (Fig. 3C), cell migration in both a wound-healing assay (Fig. 3D) and a Boyden chamber assay (Fig. 3E), and colony formation (Fig. 3F). In vivo, silencing MYB using targeted short hairpin RNA (shRNA) in PC-3M cells (Fig. 3G) injected into the prostates of athymic nude mice significantly reduced xenograft tumor growth and spontaneous metastasis in mice assessed by bioluminescence (Fig. 3H), tumor wet weight (the weight of tumors immediately after collection from euthanized mice) (Fig. 3I), and analysis of lymph nodes (Fig. 3J). IHC analysis on day 28 (the experimental endpoint) tumors treated with shMYB or control shRNA did not show differences in apoptosis; however, the apoptotic activity relative to proliferative activity, measured as the staining ratio of apoptosis marker terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) to the proliferation marker Ki67, was significantly higher in shMYB-transfected tumors than in the controls because of significantly reduced proliferation (Fig. 3K). We speculate that the low apoptosis rate in shMYB-transfected tumors may be the consequence of selection of apoptosis-resistant cells in this model.

Fig. 3 c-Myb silencing inhibits PCa growth in vitro and in vivo.

(A) MYB siRNA efficiency by qRT-PCR analysis. (B to F) Percentage of sub-G1 cells (B), cell proliferation (C), cell migration (D and E), and colony growth (F) in PCa cells untreated (Un), mock-transfected (Mock), or transfected with control (siNC) or MYB siRNA. (G) Western blotting for c-Myb in PC-3M cells transfected with one of two MYB shRNAs. (H) Representative bioluminescent images of control (shNC) or MYB knockdown PC-3M xenografts at day 21. Images are representative of six mice per group. (I) Analysis of mean tumor weight ± SD from six mice in each group at day 28. (J) Incidence of lymph node metastasis in mice bearing control (17 mice) or MYB knockdown (9 mice) xenografts. (K) TUNEL staining (AP), Ki67 staining, and their ratio in control (n = 6) and MYB knockdown xenografts (n = 5). All data are representative images or means ± SE from three experiments. *P < 0.05, Wilcoxon rank sum test.

AR- and c-Myb–regulated DDR gene signature is strongly correlated with metastasis, castration resistance, and tumor recurrence

Through microarray analysis of both siRNA- and CSS-treated PCa cell samples, we identified specific gene sets that were regulated by AR and c-Myb (Fig. 4A and table S3). Gene ontology enrichment analysis revealed a marked similarity in the biological processes regulated by AR and c-Myb (Fig. 4B) and a subset of DDR-associated genes regulated by both AR and c-Myb (Fig. 4C). Using published microarray data (27), we found that the DDR-associated genes listed in Fig. 4C have high frequencies of overexpression in human PCa patient samples, especially in metastatic PCa (Fig. 4D). These data suggested that c-Myb and/or AR activities induce the expression of DDR-associated genes in PCa and that c-Myb may substitute for AR as the dominant regulator of their common DDR target genes in castration-resistant or AR-negative PCa.

Fig. 4 Correlation analysis of the DDR gene signature.

(A) Venn diagram of genes that had altered expression in cells transfected with siAR or CSS, or siMYB. (B) Gene ontology enrichment analysis for the top five biological processes inhibited by siAR or CSS, or siMYB. (C) Heat map for DDR-associated genes that had decreased expression in cells transfected with siAR or CSS, or siMYB. (D) AR, MYB, and target gene expression in primary and metastatic PCa by the analysis of online data (27). *P < 0.05, **P < 0.01, χ2 test. (E to J) Correlation of the AR and c-Myb DDR gene signatures in (E) primary PCa versus metastatic PCa, (F to H) PCa that showed response to therapy that blocked androgen signaling, or (I) patients who showed no recurrence compared with those that did. (J) Kaplan-Meier curves comparing the survival of patients who had a higher than median signature score (yellow) with the survival of those that had a lower than median score (blue). (K) IHC analysis for DDR proteins in normal human prostate, primary PCa, or bone metastasis tissues. Scale bars, 80 μm. P values shown were derived using Kruskal-Wallis rank testing.

To test the potential clinical implications of these findings, we generated a DDR gene signature consisting of AR, MYB, and 34 AR and c-Myb co-regulated DDR-associated genes. Using six independent prostate-specific gene expression microarray data sets from the Gene Expression Omnibus (GEO) database, we calculated enrichment scores for our DDR gene signature for each sample in every data set. The results showed that the enrichment scores of our DDR gene signature positively correlated with PCa metastasis, castration resistance, tumor recurrence, and reduced overall survival in PCa patients (Fig. 4, E to J). IHC analysis of the proteins encoded by three selected DDR-associated genes—BRCA1, TOPBP1, and XRCC3—confirmed that these three were also increased at the protein level in human PCa specimens, especially in bone metastases from CRPC patients (Fig. 4K). Together, the data from these analyses suggest that our DDR gene–derived signature is significantly, positively correlated with major features of aggressive PCa, and demonstrate a promising predictive potential for clinical application.

Silencing of MYB, BRCA1, or TOPBP1 synergizes with PARP inhibitor to increase PCa cytotoxicity

To address whether silencing of MYB or selected DDR signature genes could synergize with PARP inhibition to increase cytotoxicity to PCa, we performed cell cycle analysis to find optimal combinations of DDR gene silencing and PARP inhibition. Among the DDR signature genes, we selected BRCA1 and TOPBP1 because of the synthetic lethality created by PARP1 inhibition and BRCA1/2 deficiency, and the crucial role of TOPBP1 in the activation of the ATR-Chk1 pathway. Transfection of siRNAs targeting MYB, BRCA1, or TOPBP1 into PCa cells effectively suppressed the expression of their corresponding target genes (Fig. 5A). Consistent with our microarray and qRT-PCR data (Fig. 4C and fig. S3), MYB silencing reduced the protein abundance of BRCA1 and TopBP1 (Fig. 5A). Cell cycle analysis showed that silencing MYB, BRCA1, or TOPBP1 increased the sub-G1 cell population, with an accompanying decrease in the G1 cell population (Fig. 5, B to E, and table S4). Treating cells with the PARP inhibitor olaparib (OLA) significantly reduced the percentage of G1 cells, increased that of sub-G1 cells in all PCa cell lines, and markedly increased that of G2-M cells in LNCaP and PC-3M, suggesting that OLA increased the number of DNA-damaged, arrested cells and the number of apoptotic cells. The combination of DDR gene silencing and OLA further reduced the proportion of G1 and G2 cells and further increased that of sub-G1 cells (Fig. 5, B to E, and table S4). Synergistic (more than additive) cytotoxic effects were achieved in most cases by the combination of siMYB, siBRCA1, or siTOPBP1 and OLA (Fig. 5, B to E, and table S5).

Fig. 5 DDR-associated gene silencing synergizes with PARP inhibition to increase cytotoxicity to PCa cells.

(A) Western blotting for efficiency of MYB, BRCA1, and TOPBP1 siRNAs, representative of three or more experiments. (B to E) Cell cycle analysis in PCa cells untreated, mock-transfected, or transfected with specific siRNAs (20 nM), cultured for 24 hours, and treated with vehicle (C) or OLA (O; 10 μM) for 48 hours. Data are representative profiles and means ± SD of three experiments. #Synergistic effect, summarized in tables S4 and S5.

Targeting the c-Myb–TopBP1–ATR–Chk1 pathway using the Chk1 inhibitor AZD7762 synergizes with OLA in the treatment of PCa

As noted above, TOPBP1 is one of the important targets of AR and c-Myb and is a crucial activator of the ATR-Chk1 cell cycle checkpoint and DDR pathway. Silencing TOPBP1 by siRNA markedly reduced the phosphorylation of Chk1 at both Ser317 and Ser345 (Fig. 6A). Because no specific pharmacological inhibitors currently exist to target c-Myb and TopBP1, targeting Chk1 downstream of these may be a viable alternative strategy. Cell cycle analysis showed that the Chk1/2 inhibitor AZD7762 (AZD) abrogated cell cycle checkpoint arrest, reducing the proportion of cells in G1 and G2-M and increasing the proportion in sub-G1 (Fig. 6B). The combination of AZD and OLA further reduced the proportion of G1 cells (Fig. 6B) and synergistically increased the proportion of sub-G1 cells, a measure of cytotoxicity (Fig. 6C), and DNA fragmentation (Fig. 6D), and inhibited colony growth (Fig. 6E) in both AR-positive and AR-negative PCa cells (see also table S5).

Fig. 6 Chk1 and PARP inhibitors have synergistic effects in PCa cells.

(A) Western blotting analysis for the ATR-Chk1 signaling components after TopBP1 knockdown. P, phosphorylated. Blots are representative of three or more experiments. (B and C) Representative cell cycle profiles (B) and cell cycle analysis (C) in PCa cells treated with vehicle [dimethyl sulfoxide (DMSO); C], Chk1 inhibitor AZD (A; 200 nM), OLA (O; 10 μM for VCaP and LNCaP or 5 μM for CWR22Rv1 and PC-3M), or a combination (A + O) for 48 hours. Profiles are representative and data are means ± SD from three or more experiments. (D) DNA fragmentation in cells treated and analyzed as in (C). (E) Colony formation assay in cells treated and analyzed as in (C). #Synergistic determination shown in table S5.

To assess the translational applications of this drug combination, we confirmed our in vitro findings in vivo using a VCaP orthotopic xenograft PCa model. VCaP is derived from a vertebral metastatic lesion of a patient with hormone-refractory PCa (28) and has been used in preclinical studies with PARP inhibitors (23, 29). Our in vivo data showed that OLA or AZD alone inhibited tumor growth, but the combination of AZD and OLA significantly increased this therapeutic effect (Fig. 7, A to C). IHC analysis of tumors at day 35 showed significantly reduced cell proliferation by Ki67 staining and an increased ratio of apoptotic to proliferating cells (TUNEL/Ki67) when mice were treated with the combination of AZD and OLA (Fig. 7D), indicating that the combined inhibition of PARP and Chk1 may be a more effective therapy for PCa. The low frequency of TUNEL-positive cells may be due to selection for antiapoptotic cells at the end of drug treatment.

Fig. 7 Chk1 and PARP inhibitors synergistically suppress xenograft tumor growth in vivo.

C, vehicle control; A, AZD; O, OLA; A + O, AZD + OLA. (A and B) Representative (A) and quantitative (B) bioluminescence images showing xenograft tumor growth in mice treated with single agents (C, DMSO; O, OLA; A, AZD) or combination therapy (A + O). Data are means ± SE from each experimental group (C, 11 mice; A, 10 mice; O, 10 mice; and A + O, 10 mice). *P = 0.0459. (C) Wet tumor weight in mice from each group. Data are means ± SE from the number of mice indicated in each bar. (D) IHC analysis of TUNEL and Ki67 staining and their ratio in the tumors treated with vehicle (V, n = 8), OLA (O, n = 8), AZD (A, n = 7), or A + O (n = 7). *P < 0.05. (E) Proposed AR and c-Myb co-regulated DDR signaling pathway. In AR-positive, androgen-sensitive PCa, AR and c-Myb co-regulate TOPBP1, BRCA1, and other their common DDR-associated genes, promoting ATR-Chk1 signaling and DNA repair. In AR-negative PCa or upon ADT or the impairment of AR signaling, MYB is derepressed and predominantly promotes DDR gene regulation. *Druggable therapeutic targets. Dashed outline highlights the major molecular events associated with this study.


Here, we identified that MYB is transcriptionally repressed by AR in hormone-naïve PCa cells, but is derepressed by ADT or impairment of AR signaling. We also demonstrated that MYB silencing induced PCa cell death; inhibited PCa cell proliferation, motility, and clonal growth in culture; and suppressed PCa tumor growth and spontaneous metastasis in vivo, indicating a crucial role for c-Myb in PCa cell survival and malignant progression. Our microarray data revealed that c-Myb and AR regulate markedly similar bioprocesses including a substantial subset of DDR-associated target genes. AR- and c-Myb–regulated DDR gene signature positively correlated with metastasis, castration or ADT resistance, tumor recurrence, and reduced overall survival in PCa patients. Furthermore, we identified synergistic therapeutic approaches targeting PARP and c-Myb–TopBP1–ATR–Chk1 pathway in AR inhibition–resistant PCa (Fig. 7E). Silencing MYB or its target DDR-associated genes BRCA1 or TOPBP1 generated synergistic effects with the PARP inhibitor OLA. Targeting the c-Myb–TopBp1–ATR–Chk1 pathway by using the Chk1 inhibitor AZD mimicked the synergistic effects observed with the combination of MYB, BRCA1, or TOPBP1 silencing and OLA. Although the combination of AZD and OLA was studied in pancreatic and mammary cell lines (30, 31), this study is the first in vitro and in vivo demonstration of AZD synergy with OLA in PCa, providing a strong rationale for clinical development. Our results establish a new mechanistic framework for understanding the progression of PCa to a castration- and drug-resistant state, and the roles of c-Myb and its DDR-associated target genes in this process. The DDR gene signature derived from this study may function as predictive biomarkers for the efficacy of second-generation AR inhibitors or second-line therapeutic options in patients whose PCa continues to progress on these agents.

Our finding that AR regulates MYB expression is important, given the critical role of AR in the normal prostate and in PCa and the role of c-Myb as an oncogenic transcription factor with diverse functions. We emphasize that AR suppression of c-Myb occurs only at physiological concentrations of androgen (≥1 nM) and not at concentrations observed after castration (≤10 pM). This explains why a high abundance of AR and c-Myb coexists in metastatic PCa and CRPC and together promotes PCa progression. The finding of overlapping regulatory functions of AR and c-Myb with regard to DDR-associated genes in PCa is interesting given the transition of oncogenic signaling control in PCa during the development of CRPC (2). Our data support a model in which (i) AR and c-Myb co-regulate the DDR at the transcriptional level in hormone-naïve PCa, with AR functioning as the dominant regulator in most cases, and (ii) derepression of MYB in CRPC and in AR-negative PCa increases the capacity of c-Myb to promote DDR gene expression and therefore can supplant AR for the dominant role in promoting PCa cell growth, survival, castration resistance, and drug resistance. Additional studies are warranted to test this model further and to assign a phenotypic profile(s) to the transition from AR-dominant to c-Myb–dominant regulation of DDR-associated gene expression, particularly in CRPC in which both AR and c-Myb activities may play important roles in promoting the DDR.

Another important finding in this study is the identification of TOPBP1 as one of the AR- and c-Myb–regulated DDR genes, because TopBP1 is an essential activator for the ATR-Chk1 signaling pathway that plays a critical role in both cell cycle control and DNA repair. Although c-Myb and TopBP1 are not currently druggable targets, their abundances in clinical samples may function as biomarkers to predict the therapeutic resistance to OLA and the clinical efficacy of targeting Chk1. Mechanistically, the synergy observed between AZD and OLA may be the consequence of targeting two crucial DDR pathways simultaneously. PARP inhibition induces cell death by causing accumulation of cells at G2 with unrepaired DNA lesions, whereas the added abrogation of cell cycle checkpoints by a Chk1 inhibitor pushes these cells into premature mitosis and subsequent mitotic catastrophe-induced cell death. Another possibility may be that Chk1 inhibition reduces the phosphorylation of Rad51 at Thr309 by Chk1 and results in deficient homologous recombination, as reported previously (25, 32). Although clinical trials with AZD were recently terminated because of cardiotoxicity, the scientific concept of the combination of a Chk1 inhibitor and OLA remains a viable clinical strategy, and further investigations into the mechanistic details of its synergy are warranted.

Before this study, information on the biological function of c-Myb in PCa was very limited. Just one report showed that c-Myb overexpression promoted proliferation, survival, motility, and invasion in PCa cells in vitro (14). Our study provides evidence for a substantial impact of c-Myb on the growth and metastasis of PCa in vivo. Furthermore, our study reveals a crucial role of c-Myb in the regulation of the DDR to promote DNA repair and therefore enhance cancer cell survival. During the preparation of this manuscript, two publications reported AR regulation of DDR gene expression programs to govern DNA repair and cancer cell survival. Goodwin et al. reported that AR activities are induced by DNA damage, and promote DNA repair and cell survival through the regulation of its key target gene PRKDC, encoding DNAPKcs (33). Polkinghorn et al. reported that PCa treated with ionizing radiation plus androgen demonstrated enhanced DNA repair and decreased DNA damage, whereas AR inhibition caused increased DNA damage and decreased clonogenic survival (34). Our current study supports the role of AR in the regulation of DDR and promotion of PCa cell survival. It reveals that MYB is derepressed by either ADT or impairment of the AR signaling pathway, that c-Myb and AR co-regulate a substantial set of DDR genes in PCa, and that c-Myb plays a dominant role in the maintenance of high abundances of these DDR genes to support cell survival in AR-positive, castration-resistant or AR-negative, androgen-independent PCa. Together, our data show that AR, c-Myb, and DDR are mechanistically linked and that this network reveals new or combined therapeutic targets for treating advanced PCa, regardless of AR status.


Cell lines

VCaP, LNCaP, DU145, CWR22Rv1, and PC-3M 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.

Patient-derived xenografts

The MD Anderson Cancer Center PCa PDXs were developed as previously described (35, 36) with the support of the Prostate Cancer Foundation (Santa Monica, CA) and the David H. Koch Center for Applied Research in Genitourinary Cancers at The University of Texas MD Anderson Cancer Center (Houston, TX). Tissue samples for PDX development were obtained during radical prostatectomies, bone surgeries to palliate skeletal complications, or biopsies of symptomatic lesions under approval by the Institutional Review Board of The University of Texas MD Anderson Cancer Center. The information regarding PDXs is summarized in table S2.


Twenty-seven human PCa radical prostatectomy specimens with a pathologic tumor stage II (pT2) and pathologic differentiation patterns with Gleason scores of 6 (n = 11), 7 (n = 9), 8 (n = 6), or 9 (n = 1) were obtained with informed consent from patients and used to analyze c-Myb abundance and DDR-associated gene expression. Patients had received no treatment before surgery. Additionally, nine bone biopsies from the patients who had various hormonal therapies were included (table S1). Immunostaining results were first graded according to the function of the staining intensity [from 0 (negative) to 3 (strong)] and the extent of positive staining of the cancerous area (1 = <10%, 2 = 10 to 50%, and 3 = >50%). The immunostaining grades were then categorized as immunostaining score 0 (grade 0), 1 (grades 1 to 3), 2 (grades 4 to 6), or 3 (grade >6). In addition, c-Myb (Epitomics, 1792-1) and AR (Santa Cruz, sc-816) immunostainings were also performed and scored on tissue microarray slides composed of 19 tumor xenografts generated from different histologic types of PCa (see table S2). Other antibodies used for IHC were those against BRCA1 (Biocare Medical, CM345C), XRCC3 (Abcam, ab20254), TopBP1 (Abcam, ab2402), TUNEL Apoptosis Detection Kit (Millipore, S7101), and Ki67 (Santa Cruz, sc-15402).

Probes and primers for qRT-PCR

TaqMan probe and primers were purchased from Applied Biosystems for Hu-B-actin (cat. no. 4326315E) and from Integrated DNA Technologies for MYB (forward: 5′-CATGTTCCATACCCTGTAGCG-3′, reverse: 5′-TTCTCGGTTGACATTAGGAGC-3′, probe: 5′-TTATAGTGTCTCTGAATGGCTGCGGC-3′). Primer sequences for AR- and c-Myb–regulated DDR genes are listed in table S6.

Antibodies for Western blotting analysis

Antibody against AR (sc-816) was from Santa Cruz. Antibodies against c-Myb (ab109127), BRCA1 (ab16780), and TopBP1 (ab2402) were from Abcam. Antibodies against GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (21188), ATR (2790), ATR phosphorylated at Ser428 (2853), Chk1 (2360), and Chk1 phosphorylated at Ser317 (12302) or Ser345 (2348) were from Cell Signaling Technology.

siRNA transfection

PCa cells were seeded at the desired density (VCaP, 1.0 × 106; LNCaP, 5 × 105; CWR22Rv1, 3 × 105; PC-3M, 2 × 105 in six-well plates, or 1/5 or 1/30 of those densities in 24- or 96-well plates, respectively). Cells were untransfected, mock-transfected, or transfected with 20 nM siMYB or siNC the following day. RNA and protein extracts were prepared 48 hours after transfection. siRNA sequences are listed in table S7.

Construction of MYB promoter reporters

A bacterial artificial chromosome clone (RP11-32B1) containing the human MYB locus was purchased from Children’s Hospital Oakland Research Institute BACPAC Resources. PCR was performed with RP11-32B1 as template to amplify human MYB promoter sequences (−1481 to +489) with forward primer (5′-CCCAGTCAGCAGAAGTCTCAAA-3′) and reverse primer (5′-GCAGCTACTAAACAATCCAGCA-3′). The PCR product was cloned into the pGL3-Basic Luciferase Reporter Vector (Promega) Sma I site to obtain MYB pro 1.2k-ex1-int1-Luc. Then, the 875 nucleotides from 5′ end of promoter were deleted using Sac I sites to generate c. Intron 1 was further deleted with Eco RI and Bgl II to obtain MYB pro 325nt-ex1-Luc. Exon 1 was deleted by Pst I and Eco RI from MYB pro 325nt-ex1-int1-Luc to produce MYB pro 324nt-int1-Luc. MYB pro 324nt-Luc was obtained by Pst I and Bgl II deletion of exon 1 and intron 1 MYB pro 325nt-ex1-int1-Luc. All the constructs were confirmed by DNA sequencing.

ChIP assays

ChIP assays were performed with a Millipore ChIP kit (17-295). The input and immunoprecipitated DNAs were subjected to PCR using primers corresponding to the –496 to –352 base pairs of the MYB promoter (forward: 5′-AGCGGGGTTTGCTCAGGAAA-3′; reverse: 5′-GGGTCGCCGCTCCCATT-3′). Antibodies (ChIP grade) against AR (Abcam, ab74272), LSD1 (Abcam, ab17721), trimethylated H3K4 (Abcam, ab8580), trimethylated H3K27 (Millipore, 07-449), acetylated H3K27 (Millipore, 07-360), RNA polymerase II (Santa Cruz, sc-900X), and normal rabbit immunoglobulin G (Santa Cruz, sc-2025 for mouse and sc-2027 for rabbit) were used.

Wound-healing assay

A straight longitudinal incision was made on the monolayer of cells using a pipette tip 24 hours after siRNA transfection. After the removal of existing medium, fresh medium was added and the cells were incubated for an additional 24 hours (CWR22Rv1 and PC-3M), 48 hours (LNCaP), or 72 hours (VCaP). Cells were then stained with HEMA3 (Biochemical Sciences, cat. no. 122-911), and the number of cells migrating into the clear area was counted and imaged with a microscope using NIS-Elements AR2.30 software (Nikon).

Boyden chamber assay

Cells were trypsinized 24 hours after siRNA transfection, and single-cell suspensions were prepared. Cells were seeded into each Falcon Cell Culture Insert (8.0-μm pore size, 24-well format; Becton Dickinson Labware, 353097) at the desired density (VCaP, 1.5 × 104; LNCaP, 1 × 104; CWR22Rv1, 5 × 103; PC-3M, 3 × 103) in 300 μl of serum-free medium and 700 μl of complete medium in the outer well. After 16 to 24 hours of incubation, the medium and cells inside the insert were carefully removed, and cells that had transmigrated onto the outer membrane of the insert were stained, counted, and imaged as described for the wound-healing assay.

Colony assay

For siRNA assays, PCa cells were untransfected, mock-transfected, or transfected with 5 nM siMYB or siNC. Cells were trypsinized 24 hours after siRNA transfection and reseeded into six-well plates at low density (VCAP, 1.0 × 105; LNCaP, 2 × 104; CWR22Rv1, 1 × 104; PC-3M, 5 × 103) and grown for up to 2 weeks for colony formation. Colonies were fixed with cold methanol for 30 min, stained with 0.5% crystal violet in 25% methanol for 30 min, and counted and imaged as described in wound-healing assay. For drug treatment assays, PCa cells were seeded into six-well plates at low density as described above and grown for 48 hours before treatment with AZD (100 nM for VCaP or 200 nM for LNCaP, CWR22Rv1, and PC-3M; Selleckchem), OLA (10 μM for VCaP and LNCaP, or 5 μM for CWR22Rv1 and PC-3M; Selleckchem), or a combination of AZD and OLA. Cell culture medium containing drug or vehicle control (DMSO) was renewed every 3 days. Colonies were stained, counted, and imaged as described above.

MYB shRNA xenograft tumor model

Stable PC-3M cell lines were generated by the transduction of a lentivirus expressing shMYB or shNC and a lentivirus expressing bioluminescent luciferase. The PC-3M cells (1.0 × 106) were injected into the prostates of athymic nude mice. Tumor growth was monitored weekly by bioluminescent images. Mice were sacrificed, and the wet weights (weight of tumors immediately after removal from euthanized mice to avoid moisture evaporation) of the tumors were determined 28 days after cell injection. Experiments were performed in accordance with animal care guidelines.

Gene expression microarray analysis

Significance Analysis of Microarrays software (Stanford University) was used to identify specific gene sets that are regulated by siRNAs targeting AR or MYB. Gene ontology enrichment analysis using Ingenuity Pathway Analysis (Ingenuity Systems) was used to determine AR, or c-Myb top-regulated biologic processes and AR- and/or c-Myb–regulated DDR-associated genes. Microarray data were deposited in the GEO database (accession number GSE49287).

Flow cytometry analysis

PCa cells were treated with 10 μM OLA or vehicle control DMSO 24 hours after siRNA transfection. Cells were harvested 48 hours after OLA treatment, stained with propidium iodide, and analyzed on a FACSCanto II flow cytometer (BD Biosciences). Cell cycle profiles and quantitative data were obtained using FlowJo software (Tree Star Inc.). For AZD and OLA combination experiments, PCa cells were treated with AZD (200 nM), OLA (10 μM for VCaP and LNCaP and 5 μM for CWR22Rv1 and PC-3M), a combination, or vehicle control for 48 hours before harvesting.

Correlation analysis of gene signature

Six prostate-specific gene expression microarray data sets were obtained from the GEO database. Three data sets (GSE32269, GSE2443, and GSE25136) included raw data, from which we extracted an expression profile using the RMA (robust multiarray average) method of the Bioconductor affy package (R version 2.15.2). The remaining three data sets (GSE28680, GSE6811, and GSE16560) contained expression profiles that we used directly. All expression profiles were further normalized using quartile methods, and multiple probes corresponding to the same gene were collapsed using the MaxMean method of the weighted correlation network analysis (WGCNA) package (Bioconductor). For each sample in every data set, we calculated signature scores for our DDR gene signature using the single-sample gene set enrichment analysis method in the gene set variation analysis (GSVA) package (Bioconductor). For data sets that did not contain every gene in a signature, we used a reduced signature containing all the signature genes that were present. With GSE16560, we evaluated the prognostic power of each signature by classifying patients into low- and high-scoring groups depending on the median of their signature scores, and we compared survival information between the two groups using the log-rank test.

DNA fragmentation assay

Assessment of DNA fragmentation was performed with a Cell Death Detection ELISA Kit (Roche) according to the manufacturer’s protocol.

Orthotopic xenograft PCa model for AZD and OLA combination

VCaP cells were transduced with lentivirus stably expressing luciferase. Aliquots of 3.5 × 106 VCaP-luciferase cells in 15 μl of phosphate-buffered saline (PBS) were injected directly into the right lobe of the dorsolateral prostate in athymic nude male mice (Taconic Farm) to induce orthotopic tumors. The tumors were allowed to grow for 14 days before treatment. The experimental groups received intraperitoneal injections of OLA (40 mg/kg, 5 days per week), AZD (25 mg/kg, twice daily every 3 days), or the combination of these two drugs for 35 days. Tumor size was monitored by measuring the luminescence signal using the IVIS 200 Series (PerkinElmer), mice were sacrificed, and the tumors were collected and weighed after 5 weeks of treatment.

Statistical analysis

For data with a nonnormal distribution or data with a small sample size, a nonparametric method, Wilcoxon rank sum test, was used. This covers qRT-PCR data, the luciferase assay, ChIP assay, MTS assay, wound-healing assay, Boyden chamber assay, cell cycle analysis, and DNA fragmentation assay, as well as TUNEL and Ki67 IHC analyses. Analysis of variance (ANOVA) with desired pairwise comparison was used to compare a single drug and a combination of two drugs. Synergism was determined using two-way ANOVA test (37).


Fig. S1. Correlation analysis of MYB and AR mRNA expression in PCa.

Fig. S2. MYB mRNA expression under different conditions that impair AR signaling.

Fig. S3. qRT-PCR validation of AR and c-Myb co-regulated DDR genes.

Table S1. Clinical treatment associated with PCa bone biopsies.

Table S2. PDX information.

Table S3. Genes differentially expressed by either AR or MYB knockdown.

Table S4. Combinatorial effects of DDR-associated gene silencing and PARP inhibition on PCa cells.

Table S5. Determination of synergy.

Table S6. Primer sequences.

Table S7. siRNA and shRNA sequences.

References (38, 39)


Acknowledgments: We thank S. Bronson and D. Chalaire for editing the manuscript. Funding: This work was supported in part by National Cancer Institute grant R0150588 (to T.C.T.); National Cancer Institute grant P50140388, the Prostate Cancer Specialized Program of Research Excellence at The University of Texas MD Anderson Cancer Center; the NIH through MD Anderson’s Cancer Center Support Grant, CA16672; and Tony’s Prostate Cancer Research Foundation. Author contributions: T.C.T. and L.L. conceived and designed the study and wrote the paper. L.L. prepared samples for gene expression microarray and performed Western blotting analysis and flow cytometric experiments. W.C. performed gene expression microarray analysis, gene signature correlation analysis, and flow cytometric experiments. G.Y. conducted immunohistochemistry and tissue microarray analysis. C.R. performed qRT-PCR, luciferase reporter assays, and the ChIP assay. S.P., T.K., S.K., and J.W. designed and conducted xenograft model experiments for the drug synergy studies. L.L. and J.Y. contributed to the wound-healing, Boyden chamber, and colony assays. P.K.S. conducted AR and c-Myb correlation analysis using published data and our own microarray data. S.P., H.T., M.D., and W.Z. conducted the MYB shRNA xenograft mouse model experiments. T.K. and J.W. performed DNA fragmentation assays. J.-S.L. and H.-S.L. conducted gene expression microarray experiments. B.B. performed gene signature correlation analysis and synergism determination. L.X. contributed to statistical analysis. E.E., S.N.M., A.M.A., E.M.L.N.T., P.T., and N.N. contributed to pathological analysis of human PCa samples and the establishment of PDX lines. P.G.C. contributed to data interpretation and manuscript preparation. All authors contributed to data analysis. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Microarray data are deposited in the Gene Expression Omnibus (, accession number GSE49287.
View Abstract

Navigate This Article