|Home | About | Journals | Submit | Contact Us | Français|
Transcriptional activity of the androgen receptor (AR) is crucial for growth and survival of prostate cancer even upon development of resistance to androgen ablation and antiandrogen therapies. Therefore, novel therapies that can suppress AR transcriptional activity when conventional hormone therapies fail are needed. Here, we show that histone deacetylase (HDAC) inhibitors, including SAHA (vorinostat) and LBH589, which are currently being tested in clinic, could be such a therapy. HDAC inhibitors block the AR-mediated transcriptional activation of many genes, including the TMPRSS2 gene involved in fusion with ETS family members in a majority of prostate cancers. Genetic knockdown of either HDAC1 or HDAC3 can also suppress expression of AR-regulated genes, recapitulating the effect of HDAC inhibitor treatment. Whereas HDAC inhibitor treatment can lower androgen receptor protein levels in prostate cancer cells, we show that independent of AR protein levels, HDAC inhibitors block AR activity through inhibiting the assembly of coactivator/RNA polymerase II complex after AR binds to the enhancers of target genes. Failed complex assembly is associated with a phase shift in the cyclical wave of AR recruitment that typically occurs in response to ligand treatment. HDAC inhibitors retain the ability to block AR activity in castration-resistant prostate cancer models and, therefore, merit clinical investigation in this setting. The HDAC-regulated AR target genes defined here can serve as biomarkers to ensure sufficient levels of HDAC inhibition.
Current hormone therapy for prostate cancer consists of two classes of drugs: those that lower serum testosterone and androgen receptor (AR) antagonists that target the ligand-binding domain (LBD) of the receptor. Although initially effective at blocking tumor growth, these therapies eventually fail, leading to a lethal drug-resistant stage called castration-resistant prostate cancer (CRPC). Evidence suggests that CRPC continues to depend on AR function for growth, and the progression from castration sensitive to castration-resistant state involves reactivation of AR in low androgen milieu (1). Our laboratory previously used microarray profiling of seven isogenic xenograft models to show that the transition from castration-sensitive to castration-resistant disease is associated with overexpression of AR protein, which proved necessary and sufficient to confer drug resistance (2). Expression profiling and AR mutagenesis studies suggested a mechanism whereby the modest increase in AR protein level hypersensitizes cells to residual levels of ligand remaining during hormone therapy and restores the transcription of key AR-regulated genes (ARG). This increase in AR protein can also convert the AR antagonist bicalutamide into an agonist. Therefore, novel agents that can disrupt AR function in the setting of overexpression are needed.
Transcription of AR target genes is regulated by the assembly of a multiprotein transcription factor complex. Agonists promote recruitment of AR and coactivators that have histone acetyltransferase activity to promoters of AR target genes, leading to histone acetylation and active transcription (3). In contrast, AR bound to antagonists, such as bicalutamide, recruits corepressors, such as NCoR or SMRT, that complex with histone deacetylases (HDAC) and repress gene expression (4). This and other evidence correlate histone acetylation with active gene transcription. Consistent with this model, HDAC inhibitors can relieve transcriptional repression mediated by nuclear receptors (5-7). By analogy, the HDAC inhibitor trichostatin A (TSA) has been reported to augment AR activity, as measured by androgen-dependent reporters and PSA (3, 8). However, two lines of evidence suggest that HDACs may be required for active transcription of ARGs. First, HDACs are overexpressed in prostate cancer and overexpression is associated with poor outcome (9). Second, HDAC inhibitors have greater antiproliferative activity against steroid receptor–positive prostate and breast cancer models compared with prostate and breast cancer models that are steroid receptor–negative (10-12). If HDACs function solely as repressors of hormone receptor signaling, then HDAC inhibitors should augment steroid receptor signaling and stimulate growth.
We addressed this complexity in prostate cancer by examining the effect of HDAC inhibitors on AR function. Our data show that HDAC inhibitors decrease AR protein levels by inhibiting transcription of AR without significantly affecting AR protein stability, as previously reported (13, 14). In addition, independent of their effect on AR protein levels, HDAC inhibitors directly inhibit transcription of AR target genes. Through expression profiling, we defined a subset of AR target genes (~50%) that are HDAC-dependent. The HDAC-dependent AR target genes include TMPRSS2, the 5′ partner of a series of ETS fusion genes detected in 50% of human prostate cancers (15, 16). Expression profiling of cells with knockdown of HDAC1, HDAC2, HDAC3, and HDAC8 shows that the effects of pharmacologic HDAC inhibition on AR function can be partially recapitulated by knockdown of HDAC1 and HDAC3. Although HDAC inhibitors can lower AR protein levels through inhibition of AR transcription, we show that the predominant mechanism for interference with AR function is by blockade of RNA polymerase II (Pol II) recruitment to the promoters of HDAC-dependent AR target genes.
TSA, sodium butyrate, and cycloheximide were obtained from Sigma. LBH589 was kindly given by Novartis. SAHA was obtained from CTEP. Immunoblot assays were done using the following antibodies: AR N-20 (Santa Cruz); acetyl-α-tubulin (Sigma); α-tubulin (Santa Cruz); acetyl-H3 (Upstate); HDAC1, HDAC2, and HDAC8 (Upstate); HDAC3 (Santa Cruz); β-actin AC15 (Sigma); and PSA C-19 (Santa Cruz). Quantitative reverse transcription–PCR (RT-PCR) was performed on an Eppendorf Realplex machine. Reverse transcription and Sybr Green real-time PCR were performed using reagents from Applied Biosystems.
AR2Pb-luciferase and ARE-luciferase were kindly provided by Robert Matusik (Vanderbilt) and Michael Rosenfeld (San Diego), respectively. Short hairpin RNA (shRNA) contructs against HDACs in the lentiviral pLKO.1 backbone were purchased from Open-biosystems (shHDAC1, TRCN0000004814; shHDAC2, TRCN0000004819; shHDAC3, TRCN0000004825; shHDAC8, TRCN0000004849). Control shRNA against firefly luciferase in the pLKO.1 backbone was a generous gift from David Sabatini (Whitehead Institute, Massachusetts Institute of Technology). Real-time primers against KLK2, NKX3.1, TMPRSS2, HDAC1, HDAC2, HDAC3, and HDAC8 were purchased from Superarray. Primers for PSA (F, 5′-CATCAGGAACAAAAGCGTGA-3′; R, ATATCGTAGAGCGGGTGTGG) and actin (F, 5′-TGTCACCAACTGGGACGACA; R, GGGGTGTTGAAGGTCTCAAA) were generated using PRIMER3 (17). Primers for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were 5′-GAAGGTGAAGGTCGGAGTC-3′ and 5′-GAAGATGGTGATGGGATTTC-3′. Primers for PSA enhancer were 5′-GCCTGGATCTGAGAGAGATATCATC-3′ and 5′-ACACCTTTTTTTTTCTGGATTGTTG-3 (18).
Animal studies were carried out under protocols approved by the Institutional Animal Care and Use Committee, and institutional guidelines for the proper human use of animals in research were followed. Nude females were obtained from National Cancer Institute Frederick Cancer Research and Development Center and maintained in ventilated caging. LBH589 was dissolved in 5% dextrose water.
For HDAC inhibitor studies, LNCaP cells used for the microarray study were charcoal-stripped for 5 d and then stimulated for 18 h in the presence of 10 μg/mL cycloheximide with no HDAC inhibitors and two concentrations of TSA (0.1 and 1.0 μmol/L), SAHA (0.5 and 5.0 μmol/L), and LBH589 (10 and 100 nmol/L), each with or without 1.0 nmol/L R1881. RNA was isolated using RNAeasy kit (Qiagen), labeled according to manufacturer’s instructions (Affymetrix), and hybridized to GeneChip U133A 2.0 Chips. Labeling and hybridization were performed by the Memorial Sloan-Kettering Cancer Center Genomics Core Facility. All experiments were performed in duplicate, and raw CEL data can be downloaded from National Center for Biotechnology Information Gene Expression Omnibus accession GSE9000 and GSE12438. Data analysis was performed using GeneSpring software (Agilent Technologies). Expression changes were considered significant if the absolute expression was >250 in all samples and the fold-change was >1.7. Scatter plots were generated using SigmaPlot (Systat software), and heat map was generated using HeatMap Builder (Stanford University).
For HDAC knockdown studies, LNCaP cells were infected with lentiviral shRNA against individual HDACs or control (luciferase) at multiplicity of infection of ~5, and no selection was performed. Three days after infection, cells were plated in charcoal-stripped serum, and 6 d after infection, cells were treated in duplicate with vehicle or 1 nmol/L R1881 for 18 h in the presence of cyclohexamide, as above.
LNCaP cells were grown in 15-cm plates with 5% charcoal-stripped/dextran-treated fetal bovine serum (CS-FBS) for 5 d (90% confluency) and then stimulated as indicated. At the indicated time, cells were cross-linked with 1% formaldehyde and chromatin immunoprecipitation (ChIP) was performed, as previously described, using primers previously published for the PSA enhancer (19). Quantification was performed using real-time PCR of the immunoprecipated template. AR (N-20), SRC1 (M-371), and p300 (N-20) antibodies were from Santa Cruz Biotechnology, and Pol II (8WG16) was from Covance.
Previous studies using reporter assays rather than endogenous target genes have found that HDAC inhibitors enhance AR transcriptional activity (20, 21). As expected, we also observed that TSA potentiates AR activity when assessed using transfected templates, such as a multimerized ARE reporter or the modified probasin promoter (Supplementary Fig. S1). This potentiation occurred in transient assays, wherein the reporter plasmids are episomal, or in stable transfectants, wherein the reporter is presumably integrated into chromatin. However, TSA blocked androgen-induced PSA production in the LNCaP prostate cancer cell line at the same concentration required to induce acetylation of tubulin and histone H3 (Fig. 1A). Other HDAC inhibitors, including sodium butyrate (Fig. 1B) and clinically relevant compounds SAHA and LBH589 (Fig. 1C), all inhibit R1881-mediated PSA stimulation at concentrations that stimulated histone H3 acetylation. In a separate AR-positive cell line, LAPC4, R1881 stimulated PSA transcription at ~2.6-fold and is not significantly affected by the low concentration of SAHA or LBH589. SAHA (5 μmol/L) and LBH589 (100 nmol/L), which inhibited PSA in LNCaP cells, profoundly inhibited PSA transcription to a level even below the nonstimulated state (Fig. 1D).
Because residual AR activity, despite low androgen levels, underlies castration resistance, the ability of HDAC inhibitors to suppress PSA below that in charcoal-stripped serum suggests that they may be particularly useful in CRPC. We, thus, directly asked whether HDAC inhibitors are also able to block AR function in CRPC models. LNCaP-AR cells, generated by over-expression of AR, can grow in castrate mice and are resistant to conventional antiandrogens, such as bicalutamide (2). As expected, compared with LNCaP-Neo cells, LNCaP-AR cells produce more PSA in the uninduced state (Fig. 1E, lane 1 versus lane 5) and are resistant to bicalutamide. In contrast, sodium butyrate completely inhibited PSA production in both LNCaP-Neo and LNCaP-AR cells.
Previous work has established that HDAC inhibitors impair the growth of hormone-dependent prostate cancer xenografts, such as CWR22 (11), but the effect on CRPC is unknown. CWR22Rv1 is a castration-resistant xenograft derived by selection of CWR22 in castrate mice (22). CWR22Rv1 expresses high levels of PSA at baseline, which is only modestly stimulated by R1881, indicating high basal AR activity. Yet, both SAHA and LBH589 can significantly inhibit PSA transcription well below baseline (Fig. 1F). Next, we examined the effect of LBH589 on the growth of CWR22Rv1 propagated in female nude mice. LBH589 (20 mg/d given for 5 days) completely blocked the growth of established tumors, whereas tumors in control animals increased 4-fold in size over 18 days (Fig. 1G), thereby demonstrating that HDAC inhibitors have antitumor activity against CRPC.
We next investigated the mechanism by which HDAC inhibitors interfere with AR activity. We explored three possible mechanisms: (a) increased degradation of AR by inhibition of HSP90 (14, 23), (b) inhibition of AR transcription (24), and (c) direct inhibition of transcription of AR target genes. We observed a decline in AR protein level when LNCaP cells were treated with TSA or sodium butyrate (Fig. 2A), consistent with either increased AR degradation or decreased transcription. To distinguish between the possibilities, we first measured AR mRNA and found decreased levels after TSA treatment (Fig. 2B). We then treated LNCaP cells with cycloheximide to block new protein synthesis and followed the turnover of AR over time. The rate of AR decline over 24 hours was not appreciably affected by TSA, suggesting no significant increase in the rate of degradation (Fig. 2C). Next, we used LNCaP-AR cells, in which the long terminal repeat (LTR) promoter, whose transcription is not suppressed by HDAC inhibitors, drives expression of exogenous AR. Whereas endogenous AR levels in LNCaP-Neo cells declined in response to TSA treatment, AR levels in LNCaP-AR cells did not (Fig. 2D). These data indicate that transcriptional suppression of AR, but not enhanced protein degradation, is the dominant mechanism by which AR protein levels are lowered by HDAC inhibitors. Despite the rescue of AR protein levels by exogenous AR expression in LNCaP-AR cells, sodium butyrate and TSA blocked the synthesis of PSA mRNA as efficiently as in parental LNCaP cells (Fig. 2E; see also Fig. 1E), indicating that HDAC inhibitors are also PSA expression independent from the lowering of AR mRNA and protein levels.
Having established that the activity of HDAC inhibitors on PSA expression is not solely due to lowering AR levels, we sought to characterize the effect of HDAC inhibition on AR target genes more broadly. We conducted microarray-based expression profiling studies of charcoal-stripped LNCaP cells treated with two doses of TSA (0.1 and 1 μmol/L), SAHA (0.5 and 5 μmol/L), or LBH589 (10 and 100 nmol/L) with and without androgen stimulation. The low and high concentrations inhibit PSA transcript by ~30% and 100%, respectively (Supplementary Fig. S2). Cycloheximide was included to block new protein synthesis and, thereby, bias the analysis toward primary AR target genes. In addition, cyclohexamide should negate the suppressive effect of HDAC inhibitors on AR transcription and ensure similar AR protein levels in all conditions.
We identified 159 (1.6%) genes whose expression was induced 1.7-fold by 1 nmol/L R1881, which we refer to as ARGs (Fig. 3A; Supplementary Table S1). This list shows substantial overlap with previously published gene lists in LNCaP cells (25, 26). Next, we examined the effects of HDAC inhibition on gene expression. As expected, treatment with higher dose of HDAC inhibitors induced a substantial change in expression profile, changing the expression of 23% genes over 1.7-fold (up or down in roughly equal distribution) in the case of SAHA and LBH589 and 27% in case of TSA (Fig. 3B for SAHA). Although this number is higher than quoted historically (27, 28), it is consistent with recent work using more sensitive microarray technology (29).
We then restricted analysis to the 159 ARGs defined above. Whereas high-dose SAHA treatment did not significantly affect the uninduced expression level of these genes, it inhibited R1881-induced expression by >1.7-fold in 72 of 159 (45%) of the ARGs (Fig. 3C). The majority of the remaining ARGs were affected by <1.7-fold, and only nine genes (5.6%) were induced. These percentages are substantially different from the effects seen on global gene expression.
To examine potential dose-dependent and compound-dependent effects of the three HDAC inhibitors, we generated a heat map of up-regulated ARGs ranked in order of effect by 1.0 μmol/L TSA (most inhibited to most induced; Fig. 3D) and scatter plots comparing the different conditions (Supplementary Fig. S3). Several conclusions arise from this analysis. First, the effects on AR target gene expression are consistent across all three compounds, with SAHA and LBH589 showing the greatest similarity. Second, the effects on ARGs are not seen until the higher dose. This finding is relevant because prior work has raised the possibility of a biphasic effect on PSA expression, with stimulation at low doses (3, 21) and suppression at high doses (our work). Third, the ARGs at either end of the heat map show substantial changes in gene expression and are candidates for pharmacodynamic biomarkers of HDAC inhibition in prostate cancer. HDAC-dependent expression of a subset of these genes was confirmed by quantitative RT-PCR in cells treated independently with TSA (Fig. 3E). Notably, this list includes the androgen-regulated gene TMPRSS2, which drives the expression of the oncoproteins ERG or ETV1 in human prostate cancers with TMPRSS2 translocations (16, 30).
Having shown that HDAC inhibitors block expression of many ARGs, we asked if shRNA knockdown of a single HDAC can recapitulate the transcriptional effects of pharmacologic HDAC inhibition. TSA, SAHA, and LBH589 inhibit both classes I and II HDACs, whereas sodium butyrate and MS-275 (which similarly blocks PSA expression; data not shown) are class I–specific. Furthermore, HDAC1, HDAC2, and HDAC3 can be recruited to the PSA promoter by androgen stimulation (3). Therefore, we focused our analysis on the four class I HDACs (HDAC1, HDAC2, HDAC3, and HDAC8).
LNCaP cells expressing individual shRNAs against each of these four HDACs were generated and showed selective protein and mRNA knockdown of the appropriate HDAC (Fig. 4A and Supplementary Fig. S4). The microarray-based mRNA expression profiling of HDAC knockdown cells treated with or without R1881 revealed that HDAC1 knockdown, and to a lesser degree HDAC3 knockdown, caused suppression of ARGs, whereas HDAC2 and HDAC8 knockdown caused more modest changes in ARG expression (Fig. 4B). HDAC1 knockdown suppressed 42% of ARGs by >1.7-fold and induced 8%, which is very similar to SAHA. Examination of individual ARGs that was suppressed by SAHA treatment (black columns) revealed that KLK2 and PSA were significantly suppressed by both HDAC1 and HDAC3 knockdown; NKX3.1 was suppressed by HDAC1 knockdown, whereas TMPRSS2 was unaffected (gray columns; Fig. 4C). This result suggests that some ARGs, such as KLK2 and PSA, require activity of both HDAC1 and HDAC3 for optimal expression, whereas others, such as NKX3.1, only require HDAC1. TMPRSS2, although suppressed by HDAC inhibitors, is unaffected by any single HDAC knockdown, raising the possibility of redundancy or transcriptional control by another less abundantly expressed HDAC that we have not directly tested.
Our data that HDAC inhibitors block PSA production in LNCaP-AR cells despite preserved AR protein level (Fig. 2D) and that ARGs are down-regulated in the presence of cyclohexamide indicate that HDAC inhibitors directly block transcription of ARGs. To investigate the mechanism, we first examined recruitment of AR and RNA Pol II to the AR binding site in the PSA enhancer using ChIP (31). AR was recruited to the PSA enhancer 4 hours after R1881 treatment, and this was not significantly altered by TSA. In contrast, Pol II was recruited to the PSA enhancer by R1881, and this recruitment was inhibited by TSA treatment (Fig. 5A). This further validates that HDAC inhibitors directly block transcription of PSA. In addition, this suggests that HDAC inhibitor–induced alterations of chromatin structure impair the ability of AR to recruit the transcriptional complex. Because assembly of p160 (SRC1, GRIP1, and AIBP) and p300 (p300, CREB) class of coactivators downstream of AR binding help recruit Pol II (3), we asked if these factors are properly recruited in the presence of TSA. At 4 hours after R1881 addition, recruitment of SRC1 and p300 was inhibited by TSA. This indicates that HDAC inhibitors block coactivator recruitment by AR (Fig. 5B).
The assembly of hormone receptor transcription factor complexes on chromatin is dynamic, with evidence that both estrogen receptor and AR cycle on and off their binding sites in a cyclical manner (32, 33). Therefore, we examined the effect of TSA on the time course of AR and Pol II recruitment to the PSA enhancer. AR was recruited to the PSA enhancer in a cyclic fashion (with a periodicity of 80–90 minutes) with the lagging end of each wave recruiting Pol II (Fig. 5C). Consistent with estrogen receptor activity on the pS2 promoter, the first recruitment wave was diminutive compared with subsequent ones (32, 34). In the presence of TSA, AR binding was observed throughout the time course but the phase of recruitment was dramatically offset. The first cycle of AR binding was abruptly terminated after 40 minutes, and a new cycle was observed, which was 180° out-of-phase relative to non–TSA-treated cells. Throughout the time course, this new AR cycle was unable to efficiently recruit Pol II, consistent with the decreased transcription of PSA in the presence of HDAC inhibitors.
HDACs are traditionally associated with transcriptional repression, particularly suppressing gene expression by a number of nuclear receptors (35). Hormone receptor antagonists, such as bicalutamide for AR and tamoxifen for the estrogen receptor, promote the assembly of HDAC complexes at hormone receptor binding sites to block ligand-induced gene expression. Here, we show, using genetic and pharmacologic approaches, that HDACs are also paradoxically required for activation of a substantial fraction of AR target genes including the transcriptional driver of ETS fusion mRNAs (TMPRSS2) implicated in ~70% of human prostate cancers. Our data illustrate the important clinical implications for the development of HDAC inhibitors as potential prostate cancer drugs.
Using shRNA, we showed that knockdown of individual HDACs recapitulates what is seen with treatment of HDAC inhibitors on transactivation of nearly all the androgen receptor target genes and AR mRNA levels. This was specifically observed with loss of HDAC1 and, to a lesser extent, loss of HDAC3. Indeed, knockdown of these two HDACs can account for the majority of the antiandrogen effect. However, neither loss of HDAC1 nor loss of HDAC3 phenocopies the effect of HDAC inhibitors on androgen-induced TMPRSS2 expression. This suggests that androgen regulation of TMPRSS2 is not dependent on a single class I HDAC but rather a combination of them. This question can be addressed by performing combinatorial knockdown studies of the individual histone deacetylases and ChIP on individual promoters.
Further investigation is required to discern precisely how HDAC inhibition interferes with AR complex assembly on chromatin. The cyclical nature of AR (and ER) binding to target genes is well established, including the nonproductive first wave of receptor binding that terminates in the presence of HDACs at the promoter/enhancer. Correspondingly, each wave of transcription is associated with a round of acetylation and deacetylation of histone H3 and H4. HDAC inhibitors could interfere with a proposed, albeit controversial, action of HDACs in resetting the promoter for assembly of a competent transcriptional complex (32). Our data, showing that TSA treatment shifts the phase of cyclical AR binding to the PSA enhancer, are consistent with this model. ChIP experiments indicate that HDAC inhibitor treatment rapidly and substantially enhances histone H3 acetylation at the PSA enhancer,5 indicating the constitutive presence of strong HAT activity at the loci. Therefore, one possibility is that hyperacetylated chromatin is incompatible with recruitment of coactivators and RNA Pol II.
Prior work has shown that HDAC inhibitors potentiate AR transcription, consistent with the transcriptional repression model. We reconcile our findings with these earlier reports in two ways. First, hyperactivation of AR function by HDAC inhibitors was observed using AR-dependent reporter constructs, which are unlikely to reflect the chromatinized state of endogenous genes. Second, the stimulatory effect of HDAC inhibitors on endogenous genes, such as PSA, was seen at doses (100 nmol/L TSA) below those required for achieving an increase in global histone acetylation. At this dose, we did not see a significant affect on PSA (Fig. 1A) or ARGs in expression profile (Fig. 3D). Higher doses are required to suppress AR function and may be indicative of a biphasic response to HDAC inhibition in certain contexts that could have clinical relevance if traditional end points, such as serum PSA levels, are used to monitor response to treatment. The HDAC-dependent and HDAC-repressed AR target genes identified in our gene array studies could serve as biomarkers for selecting those that ensure sufficient levels of HDAC inhibition in future clinical trials of these agents.
The most immediate clinical implication of our data is that HDAC inhibitors may have activity in prostate cancer. Conventional antiandrogens, such as bicalutamide, are highly effective as initial therapy but inevitably fail due to the emergence of drug-resistant disease. Here, we show that HDAC inhibitors remain potent inhibitors of AR function even in the setting of bicalutamide resistance and have antitumor activity in hormone refractory xenograft models.
Collectively, these data justify a careful examination of the therapeutic potential of HDAC inhibitors in prostate cancer. However, the magnitude of HDAC inhibition required to inhibit AR activity is critical because the effects we observed are clearly dose-dependent. Pharmacokinetic studies of the HDAC inhibitor SAHA suggest that the levels required for maximal AR inhibition are unlikely to be achieved using the oral regimen currently approved for lymphoma therapy (36). Indeed, phase I studies of SAHA, LBH589, and depsipeptide in CRPC, conducted by us and others, have been disappointing, with zero of eight PSA responders to LBH589 (37) and 2 of 31 PSA responders to depsipeptide (38). However, doses capable of AR inhibition should be possible with i.v. delivery or possibly through high dose of intermittent oral therapy, and the HDAC-regulated AR target genes defined here could be useful biomarkers to guide dose selection. This has led us to initiate a phase I study of intermittent i.v. LBH589.
Grant support: U.S. National Cancer Institute, University of California at Los Angeles (UCLA) Prostate Specialized Programs of Research Excellence seed grant, UCLA Medical Scientist Training Program (D.S. Welsbie), Ruth L. Kirschstein National Research Service Awards T32CA09056 (J. Xu) and 5T32CA009207 (Y. Chen), AACR–Bristol-Myers Squibb Oncology Fellowship (Y. Chen), and ASCO Young Investigator Award (Y. Chen).
C.L. Sawyers is a Doris Duke distinguished clinical scientist and an investigator of the Howard Hughes Medical Institute.
We thank Sawyers laboratory members Jennifer King and Phillip Watson for helpful discussions.
5Y. Chen, unpublished data.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Disclosure of Potential Conflicts of Interest H.I. Scher: Commercial research grant and honoraria, Novartis. N. Rosen: Honoraria, Novartis. C.L. Sawyers: Honoraria, Merck and Novartis. The other authors disclosed no potential conflicts of interest.