A Signaling Network Controlling Androgenic Repression of c-Fos Protein in Prostate Adenocarcinoma Cells*

The transcription factor c-Fos controls many important cellular processes, including cell growth and apoptosis. c-Fos expression is rapidly elevated in the prostate upon castration-mediated androgen withdrawal through an undefined mechanism. Here we show that androgens (5α-dihydrotestosterone and R1881) suppress c-Fos protein and mRNA expression induced by 12-O-tetradecanoylphorbol-13-acetate (TPA) or EGF in human prostate cancer (PCa) cell lines. Such suppression transpires through a transcriptional mechanism, predominantly at the proximal serum response element of the c-fos promoter. We show that androgen signaling suppresses TPA-induced c-Fos expression through repressing a PKC/MEK/ERK/ELK-1 signaling pathway. Moreover, our results support the hypothesis that p38MAPK, PI3K, and PKCδ are involved in the androgenic regulation of c-Fos through controlling MEK/ERK. Stable silencing of c-Fos and PKCδ with shRNAs suggests that R1881 promotes cell death induced by low-dose TPA through a mechanism that is dependent on both PKCδ and loss of c-Fos expression. Reciprocally, loss of either PKCδ or c-Fos activates p38MAPK while suppressing the activation of ERK1/2. We also provide the first demonstration that R1881 permits cell death induced by low-dose TPA in the LNCaP androgen-dependent PCa cell line and that TPA-induced cell death is independent of exogenous androgen in the castration-resistant variants of LNCaP, C4-2 and C4-2B. Acquisition of androgen-independent killing by TPA correlates with activation of p38MAPK, suppression of ERK1/2, and loss of c-Fos. These results provide new insights into androgenic control of c-Fos and use of PKC inhibitors in PCa therapy.

Androgens are critical regulators of prostate development, maintenance, growth, and function (1,2). 5␣-dihydrotestoster-one (DHT), 5 the biologically active metabolite of testosterone, mediates androgenic responses by binding to and activating the androgen receptor (AR). Early-stage metastatic prostate cancer (PCa) cells typically express elevated AR levels and show increased dependence on AR for growth and survival (3). Upon ligand binding, AR in the cytosolic compartment dissociates from the chaperones HSP70 and HSP90, allowing AR to translocate to the nucleus, where it modulates transcription by binding to the promoters of androgen-responsive genes (4). Metastatic androgen-dependent PCa cells ultimately progress to castration-resistant PCa (CRPC) through mechanisms under intense investigation (5). Many studies support the hypothesis that the development of CRPC requires AR, which is activated aberrantly by various mechanisms, including AR gene amplification, AR mutations, alteration in the expression of AR co-regulators, production of intratumoral androgens, and androgenindependent activation of AR by PI3K/Akt (6), MAPK/STAT3 (7,8), cytokines (9), and Wnt signaling (10).
c-Fos, one of the first studied proto-oncogenes (11), is an early response gene/transcriptional factor involved in the regulation of numerous genes that control growth and differentiation in many tissues, including the prostate, where it is coupled to androgenic responses (12)(13)(14). The c-Fos protein regulates gene expression as an AP-1 dimeric complex mainly with a member to the c-Jun family of transcription factors (11). Although c-Fos functions largely as a promoter of growth, cell survival, oncogenesis, tumor invasion, and metastasis, and its high expression is correlated with a poor clinical outcome (15,16), c-Fos has also been shown to be critical for certain inducers of apoptosis (14,17,18). Expression of the c-fos transcript is increased rapidly in the prostate upon castration-induced androgen withdrawal, preceding the onset of apoptosis (14). Moreover, the prostate epithelium of c-Fos-null mice is resistant to castration-induced apoptosis (14,19), implicating c-Fos in androgenic control of the normal prostatic epithelium. However, c-Fos assumes a more oncogenic role in tumors, where it is likely involved in the overexpression of TGF-␤1 and TGF-␤induced metastasis (20 -24). Despite the importance of c-Fos in the prostate, the mechanism by which androgens repress c-Fos expression remains unexplored. Also uncharted are therapeutic opportunities of manipulating such regulation in CRPC.
One of the best recognized inducers of c-Fos is 12-O-tetradecanoylphorbol-13-acetate (TPA), originally identified as a potent tumor promoter but later recognized to exert other diverse biological responses, including differentiation and apoptosis (25-31). TPA has been shown to induce apoptosis and/or growth arrest of various tumor cells, including the LNCaP androgen-dependent human PCa cell line. Phorbol ester-mediated apoptosis of androgen-sensitive PCa cells is accompanied by induced expression of c-fos (32, 33). However, the role of c-Fos in TPA-induced apoptosis in PCa remains unclear. Significantly, TPA can boost the therapeutic effects of chemotherapeutic agents such as radiation, TNF-related apoptosisinducing ligand-receptor (TRAIL-R) agonist antibodies, and Taxol both in vitro and in vivo (34 -37) and has been used in clinical trials for treatment of patients with myelocytic leukemia and hematological malignancies, although with marginal success and dose-limiting toxicity (38,39). Here we utilized TPA to investigate androgen control of c-Fos expression in prostate cancer cells to elucidate the molecular mechanism of androgenic control of c-Fos expression and to develop new insights into the therapeutic control of CRPC.
Western Blotting-Samples were analyzed by immunoblotting as described previously (41). In brief, cells were seeded at a density of 2-5 ϫ 10 5 cells/2 ml of 1 or 10% dextran-charcoal-DMEM/F12 containing 15 mM HEPES, followed by 24-to 48-h treatment with DHT, R1881, or vehicle prior to TPA treatment (10 ng/ml, 1-2 h). Following treatment, cells were lysed at 4°C with radioimmune precipitation assay buffer supplemented with protease inhibitor mixture as described before (42). Equal transfer was confirmed by staining membranes with Ponceau S.
Northern Blotting-Northern blotting analysis was performed essentially as described previously (43). In brief, 10 g of total RNA was electrophoresed, and equal loading and even transfer were assessed by visualization of the 18S rRNAs. mRNA was detected with cDNA probes labeled with [ 32 P]dCTP using the Prime-It RmT random primer kit (Stratagene). Visualized bands were scanned, and the density of the signal was measured using a PhosphorImager.
Development of the c-Fos Promoter Luciferase Reporter Construct-The c-Fos promoter region (accession no. K00650 M16287) from nucleotides Ϫ122 to Ϫ862 of the translational start site was cloned into the pGL3 basic luciferase reporter vector at XhoI and HindIII cloning sites by polymerase chain reaction amplification of HEK293 DNA by the Expand highfidelity polymerase system (Roche) using the XhoI upstream primer 5Ј ATTG CTC GAG CCC GAG GGC TGG AGG 3Ј and the HindIII downstream primer 5Ј TCT AAG CTT CTC AGT TGC TCG CTG CAG A 3Ј, ordered from Integrated DNA Technologies. The promoter construct included the proximal CRE, SRE, and c-sis-inducible element sites (44). The insertion was confirmed by DNA sequencing.
Preparation of Nuclear and Cytosolic Fractions-Nuclear and cytosolic fractions were prepared as described previously (45). In brief, cells were washed with ice-cold PBS and collected with 300 l of ice-cold buffer 1 (45), followed by centrifugation at 4000 rpm at 4°C for 5 min. Cell pellets were resuspended in 100 l of buffer 1, and cell suspensions were mixed gently by inverting tubes after the addition of 100 l of buffer 1 containing 0.15% Nonidet P-40. Samples were placed on ice for 15 min and centrifuged at 4000 rpm at 4°C for 2 min. The supernatant (cytosolic fraction) was collected, and the pellet was washed twice with 300 l of buffer 1 without detergent. The cell pellet was lysed in 100 l of buffer 2 (45), and samples were then clarified at 14,500 rpm for 20 min (4°C) to prepare the nuclear fraction. Nuclear or cytosolic fractions were quantified and subjected to Western blotting analyses.
Transient Transfection and Luciferase Assay-Cells were transfected using Invitrogen Lipofectamine Plus reagent as described before (40). Luciferase activity was measured using a Promega Dual-Luciferase assay kit and an ML3000 microtiter plate luminometer.
Gene Silencing of c-Fos and PKC␦ with Lentivirus-mediated shRNAs-We obtained Mission pLKO.1 lentiviral constructs from Sigma for shRNA gene silencing. Viral supernatants were produced in HEK293T cells transfected with pLKO.1 sh-RNA, pMDG, and pCMV-dR8. 74. LNCaP cells were transduced overnight with viral supernatant (multiplicity of infection, 0.5) in the presence of 5 g/ml protamine sulfate and, 24 h later, selected with 1.5 g/ml puromycin for 4 -5 days or until 100% death of the non-transduced cells.

Androgens Down-regulate TPA-induced c-Fos Protein
Expression in Prostate Epithelial Cells-Androgen ablation causes apoptotic regression of the prostate along with enhanced signaling by TGF-␤, a well recognized inducer of apoptosis of prostatic epithelial cells (46,47). We previously reported that two TGF-␤-induced transcription factors/protooncogenes, c-Fos and Egr-1 (48), are down-regulated by androgens in LNCaP cells (40). These findings prompted us to study the mechanism by which androgens control c-fos expression in PCa cells. We found that c-Fos is also induced rapidly (1-2 h) in LNCaP cells by TPA (10 ng/ml), and 1-2 days of pretreatment with 10 nM R1881, a synthetic stable androgen analogue, significantly suppressed TPA-induced c-Fos (Fig. 1A). DHT or R1881 also represses TPA-induced c-Fos expression in other AR-positive PCa cell lines (C4-2B, VCaP, and MDA-PCa-2b) (Fig. 1A). Total and phosphorylated c-Jun levels were also shown to be altered similarly in most of these cell lines (Fig. 1A), supporting the hypothesis that androgens robustly antagonize expression of the Fos/Jun⅐AP-1 complex. Even co-treatment with R1881 noticeably suppressed TPA-induced c-Fos and c-Jun in LNCaP cells, although suppression was more robust by 24 -48 h of pretreatment with R1881 ( Fig. 1, B and C). c-Fos inhibition occurred with as low as 0.5 nM R1881 but optimally by 2-10 nM R1881 (Fig. 1C). Therefore, androgens rapidly and robustly down-regulate c-Fos protein levels induced by TPA in various AR-positive human PCa cell lines. R1881 similarly down-regulates EGF-induced c-Fos expression, as shown in LNCaP cells (Fig. 1D) and in CWR22Rv1 cells (Fig. 1E), an AR-positive CRPC cell line derived from a primary tumor. Interestingly, R1881 suppressed TPA-induced c-Fos less robustly in CWR22Rv1 cells. We next examined whether androgen represses c-Fos through proteasomal degradation. The protea-somal inhibitor MG132, which stabilized Mcl-1 and AR, did not reverse R1881 repression of TPA-induced c-Fos (Fig. 1F), indicating that R1881 suppression of c-Fos is not through proteasomal degradation, and thus more likely to occur through repression of its induced expression.

Androgenic Control of TPA-and EGF-induced c-Fos Expression Occurs through a Transcriptional Mechanism-Consistent
with a transcriptional mechanism, R1881 (10 nM, 24 or 48 h) effectively repressed c-fos mRNA expression induced by either TPA or EGF, as demonstrated by Northern blotting, RT-PCR, and real-time quantitative PCR (Fig. 2, A-D). Using a c-Fos promoter-luciferase reporter construct comprising the TPAresponsive region (nucleotides Ϫ122 to Ϫ862 from the translational start site) (p-c-Fos-luc) transfected in LNCaP cells, we show that R1881 and DHT effectively (ϳ80%) repress TPA (10 ng/ml, 2 h)-induced c-fos promoter activity (ϳ8-fold) (Fig. 3A). Similar to TPA, EGF also enhanced c-fos promoter activity by 4-fold, and R1881 repressed this activity by 50% (Fig. 3B), implicating AR as a general suppressor of the c-fos promoter.
TPA is a potent activator of both classical and novel PKCs (49,50), some of which have been shown to induce c-fos transcription through a MEK-ERK pathway (51). We therefore investigated whether androgenic suppression of the above c-fos promoter occurred through a PKC-MEK-ERK pathway using various selective chemical inhibitors: U0126 (MEK1/2), PD98059 (MEK1), SP600125 (JNK), SB202190 (p38 MAPK ), and GF109203X (PKC). U0126, PD98059, and GF10903X completely suppressed c-fos promoter activity induced by TPA, whereas SB202190 robustly enhanced TPA-induced c-fos promoter activity (Fig. 3C). These data suggest that TPA induces c-fos promoter activity through a PKC-and MEK-dependent mechanism that is repressed by p38 MAPK . In addition, R1881 did not further inhibit TPA-induced c-fos promoter activity already inhibited by MEK inhibitors but suppressed the ability of SB202190 to enhance TPA activation of the c-fos promoter, suggesting that regulation of c-fos promoter by androgens might involve PKC, MEK, and p38 MAPK .
Key regulatory sites found in the c-fos promoter include the serum response element (SRE), c-sis inducible element, a calcium response element (CRE), and AP-1/CRE. SRE, the main element of this promoter, contains a CCA/T-6-GG sequence or CArG box that can bind to SRF and TCF/Elk-1 (44,(52)(53)(54)(55)(56). We studied the regulatory site of the c-fos promoter responsible for androgenic control using luciferase reporters containing discrete response elements found in the proximal region of this promoter, namely SRE-luc, SRF-luc, and CRE-luc. In LNCaP cells transfected with each of these constructs, we showed that TPA profoundly activated SRE-luc activity (55-fold induction), and 10 nM R1881 repressed this activation by ϳ50%, whereas TPA failed to induce both SRF-luc and CRE-luc (Fig. 3D). These data suggest that the SRE site plays a significant role in conveying the effect of both TPA and R1881 on the c-fos promoter.
To test our hypothesis that androgens suppress c-Fos expression through a mechanism requiring loss of MEK activity, we examined whether co-transfection of a MEK1 construct could reverse the ability of androgens to suppress c-fos promoter activity. Transfection with low-dose (50 ng) and high-dose (300 ng) MEK1 of this expression plasmid activated the c-fos promoter by 5-fold (similar to that induced by TPA). R1881 pretreatment was still able to (up to 35%) repress activation of this promoter at low-dose but not high dose act-MEK1 (Fig. 4B).
Furthermore, constitutively active MEK1 (act-MEK1) blocked the ability of androgens to suppress TPA-induced c-fos promoter activity (Fig. 4C). These data support our model that androgens inhibit c-Fos expression at least partially through suppressing the activation of MEK1.
Androgenic Control of PKCs-We tested whether androgens have an effect upstream of MEK1/2 or via PKCs. Among the PKC isoforms tested (PKCs ␣, ␦, ␥, ⑀, , , , and ), androgens noticeably induced expression of total PKC␣ and PKC␦ but did not appear to alter the expression of the other PKCs (Fig. 4, D and E), consistent with a previous report (57). Uniquely, here we show that androgens increased levels of P-PKC␦ (Ser-643) and P-PCK␣/␤II(Ser-638/641) but repressed TPA-induced phosphorylation/activation of PKC␦ at Thr-505. Therefore, R1881 selectively represses P-PKC␦ Thr-505 despite elevated lev- els of total PKC␦. Interestingly, under these conditions, TPA suppressed levels of total PCK␦. However, basal and R1881induced P-PKC␦ Ser-643 were spared from such loss, suggesting that phosphorylation of Ser-643 protects PKC␦ from TPA-induced loss.
Role of c-Fos Suppression by Androgen and PKC␦ in TPAinduced Cell Death-Previous studies have reported that TPA induces apoptosis in LNCaP cells through PKC␦and p38 MAPK -dependent mechanisms (58) and that androgens enhance TPA-induced cell killing by transcriptional induction of PKC␦ (57). To test the role of c-Fos in TPA-induced cell death, we stably silenced c-Fos in LNCaP by lentivirus-mediated shRNA transduction (Fig. 5A). Although silencing of c-Fos did not suppress cell growth in the absence of TPA or R1881, loss of c-Fos was permissive to cell death induced by low levels of TPA (0.3-1 ng/ml) in the absence of R1881 (Fig. 5B). Moreover, R1881 permits cell killing by low levels of TPA (Fig. 5, B and E), levels that also activate p38 MAPK , but only in the presence of R1881 (Fig. 5F). Silencing of c-Fos did not alter the expression of PKC␦ but enhanced TPA-induced P-p38 MAPK and suppressed P-ERK1/2 (Fig. 5A), implying roles for ERK1/2 and p38 MAPK in the autoinduction of c-Fos and R1881 permissiveness for TPA-induced cell death.
We next silenced PKC␦ to interrogate the role of PKC␦ in the regulation of c-Fos by androgens and in TPA-induced cell death (Fig. 5, C-E). Silencing of PKC␦ suppressed TPA-induced c-Fos and P-ERK1/2 but elevated P-p38 MAPK (Fig. 5, C  and D). Silencing of PKC␦ also suppressed basal and R1881induced cell growth and effectively reversed the permissive activity of R1881 on TPA-induced cell death (Fig. 5E). Our results support the hypothesis that c-Fos, p38 MAPK , and ERK1/2 are involved in the ability of R1881 to permit TPA (0.62-1.2 ng/ml)-induced cell death, consistent with the regulation of those proteins at the above TPA doses (Fig. 5F). We therefore used 2.5 ng/ml TPA for further analyses on the basis of these dose-response results.
Cross-talk between MEK/ERK, p38 MAPK , and PI3K/Akt in TPA-induced c-Fos Expression-The opposing action of R1881 on activation of ERK1/2 and p38 MAPK by TPA is consistent with the possibility that p38 MAPK and ERK1/2 oppose the activation of each other by TPA. To test this possibility, we studied the effect of selective inhibitors of p38 MAPK and ERK1/2 on their  respective activation following 2 h of treatment with 2.5 ng/ml TPA prior to a 24 h pretreatment with 1 nM R1881 (Fig. 6, A and B). The p38 MAPK inhibitor SB203580 enhanced levels of P-ERK1/2 and c-Fos induced by TPA, but only in the absence of R1881. Interestingly, in this experiment, 24-h treatment with 1 nM R1881 alone robustly suppressed TPAinduced c-Fos but did not noticeably suppress P-ERK1/2 (Fig. 6A, first through fourth lanes). However, SB203580 enabled suppression of P-ERK1/2 by R1881 and enhanced suppression of c-Fos by R1881. These results suggest that suppression of P-ERK1/2 at 24 h of R1881 treatment is constrained by p38 MAPK and that R1881 represses c-Fos expression both upstream and downstream of ERK1/2. Consistent with a previous report (58), TPA represses the activation of Akt in LNCaP cells (Fig. 6A). SB203580 reversed loss of P-Akt Thr-308 by TPA but not by R1881 (Fig. 6A), suggesting that p38 MAPK mediates the suppression of Akt activity by TPA but not by R1881.
Consistent with ERK1/2 as a mediator of TPA-or EGF-induced c-Fos, the highly specific and potent MEK1/2 inhibitor selumetinib completely blocked TPA-induced c-Fos (Fig. 6, B and C). Selumetinib slightly suppressed P-p38 MAPK induced by TPA but enhanced P-p38 MAPK induced by R1881 alone or by TPA ϩ R1881. On the other hand, the highly potent and specific PI3K inhibitor ZSTK474, which fully blocks the activation of Akt (P-Akt Thr-308 ), significantly suppressed c-Fos induced by either TPA or EGF, suppressed P-ERK induced by TPA (but not by EGF), and also reversed the suppressive effect of R1881 on TPA-induced c-Fos expression (Fig. 6, B and C). Moreover, ZSTK474 robustly enhanced basal levels of P-p38 MAPK , suggesting that AR enhances TPA-activated p38 MAPK and cell death by suppressing PI3K. This suggests that R1881 may suppress c-Fos also in part through inactivation of PI3K. Next, in a subsequent experiment we tested whether these effects of ZSTK474 were through suppression of Akt kinase by using the selective Akt kinase inhibitor MK2206. In contrast to ZSTK474, MK2206 did not suppress c-Fos expression, although it partially reversed the ability of R1881 to suppress expression of c-Fos (Fig. 6D), suggesting that Akt kinase is involved in the androgenic suppression of c-Fos expression but not in the expression of c-Fos induced by TPA. We also conducted a time course experiment to define temporal changes in the activation of those kinases following TPA treatment and their suppression by R1881 (data not shown). TPA activated ERK1/2 and suppressed P-Akt Thr-308 as early as 5-15 min, whereas activation of p38 MAPK did not appear until 30 min, and induced expression of c-Fos did not occur until 60 min. R1881 effectively suppressed c-Fos activation at both time points. R1881 altered the magnitude of such changes but not the time of onset or the dose response of TPA (data not shown). Interestingly, at 22 h of R1881 treatment, 0.1 nM R1881 more effectively suppressed TPA-induced c-Fos and P-ERK1/2 expression than did Ն0.2 nM R1881, the latter corresponding to the dose of R1881 that starts to enhance TPA-induced P-p38 MAPK but not induce expression of prostate-specific antigen (Fig. 6E). Together, these data support a signaling model (Fig. 6F) in which TPA induces c-Fos expression through a PKC/MEK/ERK-dependent pathway and androgens repress c-Fos expression by two mechanisms: one involving suppression of ERK1/2 via activation of p38 MAPK (by AR-mediated suppression of PI3K) and the other downstream of ERK1/2 at the level of the c-Fos promoter.
Androgenic Control of the c-fos Promoter Downstream of ERK-Thus far, our data support the hypothesis that MEK/ERK signaling plays critical roles in the induction of c-Fos by TPA or EGF and that the androgenic suppression of c-Fos occurs both upstream (Ͼ24 h of R1881) and downstream (Յ24 h of R1881) of MEK/ERK. To define the mechanism by which androgens may suppress c-Fos without suppressing P-ERK levels, we first examined whether R1881 represses the nuclear transport of P-ERK. To do so, LNCaP cells were treated in the presence or absence of R1881 for 22 h before a 2-h treatment with or without TPA (2.5 ng/ml), and nuclear and cytosolic fractions were prepared and subjected to Western blotting. Although R1881 suppressed the TPA-induced cytosolic P-ERK, it enhanced rather than repressed the nuclear transport of P-ERK, in contrast to repression of nuclear c-Fos (Fig. 7A).  A). Cell lysates were immunoprecipitated with rabbit anti-AR (catalog no. sc-816) or control rabbit IgG and immunoblotted with mouse anti-AR (catalog no. sc-7305), anti-ELK-1 (catalog no. sc-56896), and anti-ERK1/2 (catalog no. 9107) antibodies. D, nuclear lysates of LNCaP cells (pretreated with or without 10 nM R1881 for 46 h, followed by 2 h with or without 10 ng/ml TPA) were subjected to EMSA using radiolabeled WT SRE and mutant SRE containing WT TCF-dimerized oligonucleotides. The DNA-protein complex (complex 1) was supershifted by treatment of the dimerized oligos with 1 g of anti-SRF IgG (catalog no. sc-335) 30 min prior to electrophoresis. exp, exposure.
A downstream target of ERK is ELK-1, which directly binds to the TCF response element, thereby recruiting SRF to activate the c-Fos promoter (51,59). A 22-h treatment of LNCaP cells with R1881 significantly repressed TPA-induced phosphorylation of ELK-1 but did not noticeably alter the expression of total ELK-1, total SRF, or P-SRF (Fig. 7B). These data suggest that R1881 directly targets ELK-1 to inhibit its interaction with the c-Fos promoter. ELK-1 has been recently reported to affect R1881 transcriptional responses by a direct physical interaction with AR (60). Consistent with this interaction, we showed that R1881 enhances binding of AR to ELK-1 but not to ERK1/2 in TPA-treated LNCaP cells (Fig. 7C).
EMSA was employed to study the binding of nuclear proteins from LNCaP cells (treated with or without R1881 (46 h) before a Ϯ2 h TPA treatment) to a minimal wild-type SRE oligonucleotide (SRE(WT)) versus a mutant SRE (SRE(MT)) in which the SRF response element, but not the TCF response element, was mutated. Nuclear lysates of control LNCaP cells retarded the migration of SRE(WT), resulting in two discrete complexes, designated C1 and C2 (Fig. 7D). C1, but not C2, was lost in SRE(MT) with an intact TCF, consistent with the association of ELK-1 with TCF in C2 (61). We tested the occupancy of SRE with SRF, which forms a ternary complex with ELK-1 on the c-Fos promoter (51,56), by treating nuclear lysates with 1 g of anti-SRF 30 min before electrophoresis. As shown in Fig. 7D, C1, but not C2, was supershifted by anti-SRF, indicative of the selective occupancy of SRF in C1. The intensity of C1 was marginally weaker and stronger in cells treated with TPA and R1881, respectively, suggesting that activation of SRE by TPA may involve the release of a co-repressor from an already formed SRF complex rather than the formation of an SRF complex and that R1881 may interfere with the release of this corepressor. On the other hand, C2 was significantly suppressed by R1881, consistent with our model that R1881 also inhibits the association of ELK-1 with TCF. These results suggest that R1881 suppresses the c-fos promoter by intercepting the binding of ELK-1 to TCF without a concomitant loss in the recruitment of SRF to SRE.

Discussion
Our data support the hypothesis that androgens repress c-Fos transcription through two basic mechanisms in PCa cell lines. One mechanism is via the PKC/MEK/ERK pathway and requires Ͼ24 h of androgen stimulation. A second, more rapid mechanism is downstream of ERK, involving ELK-1 and its interaction with AR (59). To the best of our knowledge, the mechanism by which androgens control c-Fos expression remained unexplored. Although the induction of c-Fos expression by either TPA or EGF is largely driven by activation of the PKC/MEK/ERK pathway (Figs. 3C, 4A, and 6C), our data support the hypothesis that PI3K also contributes to this induction (Fig. 6, B and C). Our study is also the first demonstration that both TPA and R1881, which cooperatively inhibit the activation of PI3K (as indicated by phosphorylation of Akt at Thr-308), synergistically activate p38 MAPK (Fig. 6B) and the resulting p38 MAPK also represses c-Fos expression (Fig. 6F).
Intriguingly, Fig. 6D supports the hypothesis that Akt kinase is involved in the mechanism by which AR suppresses c-Fos expression without controlling TPA-induced expression of c-Fos. It is likely that a direct interaction of Akt with AR is involved in this androgenic suppression of c-Fos transcription. Akt was previously shown to interact with and phosphorylate AR at Ser-210 and Ser-790 (62). Whether the dependence on Akt for androgenic suppression of c-Fos requires the phosphorylation of AR at those sites remains to be defined. Also unanswered is how ZSTK474 suppresses c-Fos expression (Fig. 6, B and D) and how R1881 partially reverses the ability of ZSTK474 to suppress c-Fos or P-ERK1/2 levels. At first glance, it would appear that ZSTK474 suppresses c-Fos expression through inhibiting ERK1/2. However, ZSTK474 suppresses EGF-induced c-Fos but not EGF-induced P-ERK1/2 (Fig. 6C), suggesting that PI3K may be involved in the induction of c-Fos through blocking P-p38 MAPK , which we suggest functions as a suppressor of c-Fos through inhibiting ERK1/2. However, the inefficiency of ZSTK474 to suppress EGF-induced c-Fos (Fig. 6C) suggests that p38 MAPK may suppress c-Fos levels through an ERK1/2-independent mechanism as well as an ERK1/2-dependent one.
Our observation that a 24-h treatment of PCa cells with 1 nM R1881 (without SB203580) far more robustly suppressed expression of c-Fos than of P-ERK1/2 (Fig. 6A) suggests that R1881 also regulates c-Fos expression through a step downstream of ERK1/2. In contrast, the same treatment with R1881 robustly suppressed the activation of the ERK1/2 substrate ELK-1, which binds to the TCF element of the c-Fos promoter, suggesting that AR suppresses c-Fos also through inhibiting ELK-1. This may occur through binding of ELK-1 to AR (60), leading to sequestration of ELK-1 from interacting with ERK1/2 or/and from binding to TCF. The latter is supported by our EMSA results (Fig. 7D), which show that R1881 robustly suppresses formation of the TCF complex. Interestingly, SRE(MT) containing intact TCF more effectively bound to C2 than did SRE(WT), suggesting that R1881 represses the formation of a TCF complex through alterations in the adjacent SRF complex.
Of note, proteasomal inhibition by MG132 did not stabilize TPA-induced c-Fos expression in LNCaP cells (Fig. 1E), in contrast with previous reports showing that c-Fos levels are robustly controlled by proteasomal degradation in other systems (63)(64)(65). Instead, we show that MG132 suppresses TPAinduced c-Fos expression while stabilizing Mcl-1 and AR (Fig.  1E). Thus, MG312 may suppress TPA-induced c-Fos expression by stabilizing a transcription/translational suppressor of c-Fos, such as AR. However, MG312 was previously reported to block AR transcriptional responses in LNCaP cells through suppressing the nuclear translocation of AR or/and interaction of AR with transcriptional co-activators (66). In our MG132 experiment, cells were pretreated with R1881 before treatment with MG132, allowing AR to first translocate to the nucleus. Thus, the AR co-regulator that may be involved in the suppression of c-Fos is unlikely to be significantly suppressed by MG132. Otherwise, a non-genomic mechanism of AR signaling (67,68) would need to be invoked for R1881-driven suppression of c-Fos.
Our data showing that androgens suppress the expression of both c-Fos and c-Jun support the hypothesis that AR is a gen-eral antagonist of AP-1 activity in prostate cancer cells. Previous studies showed AP-1 is also a transcription repressor of AR activity (69 -71). Although not discussed, Fig. 1D in one study (71) showed a modest suppression of TPA-induced c-Jun expression in LNCaP cells; however, TPA-induction of c-Fos was not noticeably altered. While the basis for differences between our and their results are not clear, one marked contributing factor may be differences in growth media used: TCM TM serum replacement was used in their study in place of the charcoal-stripped FBS used in our study. It is feasible that either a growth factor activity in TCM TM interferes with androgenic suppression of c-Fos or that a factor in charcoal-stripped FBS may be necessary for this androgenic response.
Androgen withdrawal-induced apoptosis of prostate epithelial cells has been associated with the induction of c-Fos, supporting a role of c-Fos as a regulator of apoptosis following  Fig. 6, B and E. D, androgen-independent killing of C4-2 and C4-2B cells by low-dose TPA correlates with androgen-independent activation of p38 MAPK and reduced activation of ERK1/2 by TPA, as assessed by Western blotting and growth conditions described in Fig. 6A. E, TPA failed to kill RWPE-1 and DU145 cells but suppressed R1881-induced growth in VCaP, MDA-PCa-2b, and CWR22Rv1 cells assayed as in A. F, Western blotting analysis of the expression of AR in CWR22Rv1 cells treated with or without R1881, TPA, or EGF for 3 days. Error bars represent the mean Ϯ S.E. of triplicate to six replicates. p Values were calculated by Student's t test (two-tailed) and two-way analysis of variance). Data are representative of at least two independent experiments. #, p Ͻ 0.05; *, p Ͻ 0.01; **, p Ͻ 0.001. androgen withdrawal (14). Castration-induced apoptosis is likely to be trigged by TGF-␤1, which is transcriptionally induced by c-Fos (21, 33). Previous studies have reported that TPA induces apoptosis of LNCaP cells through induced expression of PKC␦ and PKC␦-dependent activation of p38 MAPK (57,58). In contrast, our PKC␦ shRNA data support the hypothesis that p38 MAPK is suppressed rather than activated by PKC␦. Moreover, to the best of our knowledge, we report the first evidence that activation of p38 MAPK by TPA ϩ R1881 represses ERK1/2. Additionally, our data unveil further complexity to this, since TPA promotes loss of total PKC␦, and R1881 represses TPA-induced phosphorylation of PKC␦ at Thr-505 but elevates the levels of PKC␦ phosphorylated at Ser-643 (Fig.  4D). The roles of these PKC␦ modifications in androgenic responses and TPA-induced death remain to be explored.
We demonstrate for the first time that, although low-dose TPA promotes the growth of LNCaP cells under androgendepleted conditions, the same levels of TPA promote LNCaP cell death in the presence of R1881 (Fig. 5, B and E). Previous studies likely missed this permissive effect of R1881 due to the use of Ͼ50-fold molar excess of TPA (57,72). Silencing PKC␦ promotes cell growth (Fig. 5E) rather than inducing death of LNCaP by low-dose TPA, suggesting that R1881 permits TPAinduced apoptosis not simply by elevating the expression of PKC␦ but by the "switching" ability of TPA from promoting growth to promoting death. Additionally, the ability of PKC␦ to repress and enhance TPA-induced P-p38 MAPK and P-ERK1/2, respectively (Fig. 5D), is consistent with the cell survival effect of PKC␦ at low-dose TPA (Figs. 5, B and E, and 8A) under androgen-depleted conditions. Our results suggest that suppression of c-Fos and ERK1/2 and induction of p38 MAPK contribute to the apparent switch of PKC␦ function. We propose that R1881 enhances TPA-induced apoptosis through suppressing the expression of c-Fos and ERK (which function as survival factors in this context) and boosts the activation of p38 MAPK (involved in TPA-induced apoptosis of LNCaP cells (58)). These mechanisms are interdependent, since silencing c-Fos augments TPA-induced activation of p38 MAPK (Fig. 5A), and suppression of p38 MAPK activates ERK and induces c-Fos expression (Fig. 6A). How p38 MAPK represses TPA-induced ERK activation remains to be defined. Moreover, how suppression of PI3K (by ZSTK474) activates p38 MAPK in these cells also awaits further investigation.
Broadly speaking, TPA concomitantly signals the activation of both cell death and survival/growth signaling. The cell death signals are mediated through PKC␦ and p38 MAPK , and the survival signals are mediated through activation of MEK/ERK/c-Fos signaling (Fig. 6F). The effect of PKC␣, which is induced by R1881 (Fig. 4D), on androgenic suppression of MEK/ERK activity and c-Fos levels awaits further work.
Significant effort has been focused on the use of PKC modulators in cancer therapy, with a number of new agents under investigation (73). Our data showing that R1881 is permissive to the induction of cell death by very low levels of TPA offers a potentially selective and not yet not fully explored approach for targeting CRPC in which AR is overactivated. Consistent with this, TPA kills the CRPC variants of LNCaP (C4-2 and C4-2B) in the absence of exogenous androgens (Fig. 8). Our data suggest that this potentiation occurs through the ability of AR to modulate the activation of PKC␦, p38 MAPK , MEK/ERK, Akt and expression of c-Fos (Figs. 4 -6). Also, loss of c-Fos and enhanced activation of p38 MAPK may enhance the cytostatic/ cytotoxic responses of TPA in CRPC and the androgen-independent killing of CRPC cells (Fig. 8D). Moreover, it is likely that an added benefit for killing CRPC can also be achieved by independently suppressing the prosurvival activity of MEK/ ERK/c-Fos induced by TPA.
Under the above conditions, TPA is unable to suppress the growth of AR-negative tumorigenic (DU145) and non-tumorigenic (RWPE-1) prostate cells (Fig. 8E), consistent with selectivity of growth suppression/cell killing by TPA on AR-positive PCa cells. Similar to C4-2 and C4-2B cells, we found that TPA also suppresses the basal and R1881-induced growth of the ARpositive VCaP line (Fig. 8E). Two other AR-positive CRPC cell lines tested, MDA-PCa-2b and CWR22Rv1, similar to LNCaP cells, were not growth-suppressed/killed by TPA under androgen-depleted conditions, but TPA blocked their R1881-induced growth (similar to VCaP). However, unlike the LNCaP line, R1881 did not promote their death by TPA (Fig. 8E). One contributing factor for lack of the permissive ability of R1881 to promote TPA-induced death of VCaP and CWR22Rv1 may be the relatively weak suppression of c-Fos (induced by TPA) in VCaP and CWR22Rv1 cells compared with LNCaP (Fig. 1, A  and E). Taken together, these data suggest that TPA suppresses AR-positive PCa cells by suppressing AR-dependent growth as well as promoting TPA-dependent death through an AR-dependent mechanism. Our recent data on CWR22Rv1 cells suggest that TPA may suppress R1881-induced growth by suppressing AR levels (Fig. 8F), the mechanism of which remains to be explored.
While TPA has been used in a limited number of clinical trials, and shown to be effective in treating chronic myelogenous leukemia when combined with vitamin D3 or cytosine arabinoside in a clinical trial (74), TPA has not been used in clinical studies of PCa. Our cell culture studies show that well tolerated concentrations of TPA found in the blood of patients administrated TPA (37) are sufficient for R1881 to permit TPAinduced death of certain AR-positive PCa cells (Fig. 5E). Our results therefore warrant further investigation into the potential use of low-dose TPA and other PKC modulators in the therapy of advanced CRPC in patients who display elevated AR activity and failed treatment with second-generation anti-androgens.
Author Contributions-E. S., K. S., K. L. B., S. L. C., and D. D. helped with performing the experiments and acquiring and plotting/analyzing data. K. S. helped with generating the initial data and the first draft of the manuscript, and E S. was involved in generating most of the data in the final draft. H. Y. K. provided intellectual input and guidance and edited the final manuscript. H. W. contributed to the overall methodology. D. D. contributed to the initiation of the study, overall study design and methodology, and analysis and interpretation of the results and drafted and edited the manuscript during all phases of the work. The final manuscript was approved by all authors.