ROCK Mediates Phorbol Ester-induced Apoptosis in Prostate Cancer Cells via p21Cip1 Up-regulation and JNK*

It is established that androgen-dependent prostate cancer cells undergo apoptosis upon treatment with phorbol esters and related analogs, an effect primarily mediated by PKCδ. Treatment of LNCaP prostate cancer cells with phorbol 12-myristate 13-acetate (PMA) causes a strong and sustained activation of RhoA and its downstream effector ROCK (Rho kinase) as well as the formation of stress fibers. These effects are impaired in cells subjected to PKCδ RNA interference depletion. Functional studies revealed that expression of a dominant negative RhoA mutant or treatment with the ROCK inhibitor Y-27632 inhibits the apoptotic effect of PMA in LNCaP cells. Remarkably, the cytoskeleton inhibitors cytochalasin B and blebbistatin blocked not only PMA-induced apoptosis but also the activation of JNK, a mediator of the cell death effect by the phorbol ester. In addition, we found that up-regulation of the cell cycle inhibitor p21Cip1 is required for PMA-induced apoptosis and that inhibitors of ROCK or the cytoskeleton organization prevent p21Cip1 induction. Real time PCR analysis and reporter gene assay revealed that PMA induces p21Cip1 transcriptionally in a ROCK- and cytoskeleton-dependent manner. p21Cip1 promoter analysis revealed that PMA induction is dependent on Sp1 elements in the p21Cip1 promoter but independent of p53. Taken together, our studies implicate ROCK-mediated up-regulation of p21Cip1 and the cytoskeleton in PKCδ-dependent apoptosis in prostate cancer cells.

tional PKCs and novel PKCs are the target for the phorbol esters, natural products that mimic the action of the lipid second messenger diacylglycerol (1,2). Despite their well characterized tumor promoter activity, phorbol esters cause dissimilar effects, since they can either stimulate proliferation and survival or, conversely, induce cell growth arrest or trigger apoptotic cell death, depending on the cell type (2,3). Such diversity relates primarily to the differential expression of PKC isozymes according to cell type as well as to the great divergence in the signaling events modulated by individual PKCs. One of the key PKC isozymes implicated in negative growth regulation is PKC␦. Work from several laboratories, including ours, established that PKC␦ modulates the transition from G 1 to S phase of the cell cycle by controlling the phosphorylation status of retinoblastoma (4 -6). In bronchoalveolar adenocarcinoma cells, activation of PKC␦ in early G 1 leads to G 1 /S arrest through the induction of p21 Cip1 at a transcriptional level (4).
Among the few cell types that undergo apoptosis in response to phorbol esters, androgen-responsive prostate cancer cells have been one of the best characterized models. Phorbol 12-myristate 13-acetate (PMA) triggers an apoptotic response in androgen-dependent prostate cancer cells, including LNCaP, C4-2, and CWR22-Rv1 cells (7)(8)(9). The mechanisms underlying the cell death effect of phorbol esters in prostate cancer cells are only partially understood, but they seem to involve the p21 Cip1 /retinoblastoma pathway (10). Our previous studies established that this effect is primarily mediated by PKC␦, and subsequent analysis revealed that this kinase promotes the activation of the extrinsic apoptotic cascade via an autocrine mechanism. PMA promotes the secretion of death factors from LNCaP cells via PKC␦, including TNF␣ and TRAIL, and the released factors promote cell death via activation of JNK and p38 MAPK cascades (11,12).
The mammalian Rho GTPases comprise Ͼ20 proteins, among which Rac1, Cdc42, and RhoA have been the most widely studied. These small G-proteins have been established as important mediators of receptor signaling and control a variety of cellular functions related to cell division and morphology. Upon receptor activation, Rho GTPases dissociate from Rho guanine nucleotide dissociation inhibitors (Rho-GDIs), allowing Rho guanine nucleotide exchange factors (Rho-GEFs) to switch GDP by GTP and Rho activation (13). Members of the Rho family were originally established as key regulators of cytoskeletal organization in response to extracellular growth factors. Studies over the past few years have revealed that Rho GTPases also play crucial roles in diverse cellular events, such as transcriptional regulation, cell cycle control, endocytosis, differentiation, and apoptosis (13,14). Recently, growing attention has been drawn toward the emerging role of the cytoskeleton in the modulation of apoptosis. RhoA, predominantly through its effectors ROCKI and ROCKII serine/threonine kinases, regulates the phosphorylation of multiple downstream targets, including myosin light chain and LIM kinases (15,16), which control actin cytoskeleton assembly and cell contractility. It has been shown that caspase-3-mediated ROCKI activation is both necessary and sufficient for the formation of membrane blebs and nuclear disintegration in apoptotic cells (17,18). In some cell types, ROCK is involved in the intracellular signaling that initiates apoptosis, such as caspase-8, caspase-10, and caspase-3 activation (19) or modulates the transcription of the proapoptotic proteins, such as Bax (20).
Emerging evidence implicated Rho GTPases as mediators of PKC signaling. For example, the decreased invasiveness of PKC⑀-depleted head and neck squamous carcinoma cells is associated with a corresponding inactivation of RhoA and RhoC GTPases (21). PKCs can modulate Rho function through either Rho-GEF activation or Rho-GDI inactivation (22,23) or by direct association to Rho (24). However, it is not known whether Rho or its downstream effectors are implicated in PMA-induced apoptosis in prostate cancer cells.
In the present study, we show that PMA causes a strong and sustained activation of Rho and the formation of stress fibers via PKC␦. Notably, inhibition of Rho or its downstream effector ROCK impairs apoptosis triggered by PKC activation, suggesting a crucial role for Rho and ROCK downstream of PKC␦. We also found that ROCK mediates the induction of p21 Cip1 , a necessary event for apoptosis induced by PMA. All of these effects depend on the integrity of the cytoskeleton, since agents that disrupt cytoskeleton organization markedly impaired apoptosis, JNK activation, and p21 Cip1 induction by PMA.
Adenoviral Infections-LNCaP cells in 6-well plates (ϳ70% confluence) growing in RPMI 1640 medium supplemented with 2% fetal bovine serum were infected with an adenovirus (AdV) for a dominant negative RhoA mutant N19-RhoA or a GFP control AdV for 14 h at a multiplicity of infection of 10 plaque-forming units/cell. After removal of the AdV by extensive washing, cells were incubated for an additional 24 h in RPMI 1640 medium supplemented with 10% fetal bovine serum. Expression of the recombinant protein remained stable throughout the duration of the experiment, as detected by Western blotting (data not shown).
Western Blot Analysis-Cells were harvested into lysis buffer containing 50 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, and 5% ␤-mercaptoethanol. Protein determinations were performed with the Bio-Rad Dc protein assay according to the instructions provided by the manufacturer. Equal amounts of protein (20 g/lane) were subject to SDS-PAGE and transferred to polyvinylidene difluoride membranes. Membranes were blocked with 5% milk or 5% bovine serum albumin in 0.05% Tween 20, phosphate-buffered saline and then incubated with the first antibody overnight at 4°C. Membranes were washed three times with 0.05% Tween 20 in PBS and then incubated with anti-mouse or anti-rabbit secondary antibodies conjugated to horseradish peroxidase (1:3000; Bio-Rad). Bands were visualized with the ECL Western blotting detection system.
Apoptosis Assays-The incidence of apoptosis was determined as we previously described (7). Briefly, cells were stained with 4Ј,6diamidino-2-phenylindole (Sigma). Cells were trypsinized, mounted on glass slides, fixed in 70% ethanol, and then stained for 20 min with 1 mg/ml 4Ј,6-diamidino-2-phenylindole. Apoptosis was characterized by chromatin condensation and fragmentation when examined by fluorescence microscopy. The incidence of apoptosis in each preparation was analyzed by counting ϳ300 cells.
Real Time PCR-RNA was extracted using TRIzol (Invitrogen). Two g of RNA/sample were reverse transcribed using the First-Strand cDNA synthesis kit (Amersham Biosciences) following the instructions provided by the manufacturer. PCR primers and fluorogenic probes for p21 Cip1 were purchased from Applied Biosystems. The probes were 5Ј-end-labeled with 6-carboxyfluorescein. Each PCR amplification was performed in a total volume of 12.5 l, containing 6.25 l of 2ϫ TaqMan Universal PCR Master Mix (Applied Biosystems), commercial target primers (300 nM), the fluorescent probe (200 nM), and 1 l of cDNA, using an ABI PRISM 7700 detection system. PCR product formation was continuously monitored using Sequence Detection System software version 1.7 (Applied Biosystems). The 6-carboxyfluorescein signal was normalized to endogenous 18 S RNA.
Plasmid Transfections and Promoter Analyses-LNCaP cells (1.5 ϫ 10 6 cells) were transfected with 2 g of p21 Cip1 firefly luciferase reporter vectors (25) using the Amaxa Nucleofector. A Renilla luciferase expression vector (100 ng, pRL-TK; Promega, Madison, WI) was co-transfected for normalization of transfection efficiency. Forty-eight h after transfection, cells were stimulated with PMA and lysed. Cell extracts were subject to luciferase determination using the Dual-Luciferase reporter assay system (Promega).
Confocal Microscopy and Phalloidin Staining-Cells were washed twice with PBS, fixed in 4% paraformaldehyde in PBS for 10 min, and washed briefly with PBS. For phalloidin staining, cells were blocked in 5% bovine serum albumin in PBS for 30 min. Cells were stained with phalloidin (Molecule Probes) in PBS containing 1% bovine serum albumin (30 min, room temperature) and then stained with 4Ј,6Ј-diamidino-2-phenylindole (1 g/ml, 20 min, 4°C). For the determination of GFP-PKC␦ localization, LNCaP cells were transfected with pEGFP-PKC␦ plasmid using the Amaxa Nucleofector and 48 h later treated with PMA. In all cases, coverslides were mounted with Fluoromount-G (SouthernBiotech) and visualized with a laser-scanning fluorescence microscope (LSM 410 or 510; Carl Zeiss).
ROCK Kinase Assays-ROCK kinase activity was measured by two approaches. First, we determined phosphorylation of the endogenous ROCK substrate MYPT1 in total cell extracts using a rabbit anti-phospho-Thr 850 -MYPT1 antibody (27). Second, we used the ROCK Activity Immunoblot Kit (Cell Biolabs, Inc.). Briefly, cells were washed twice with ice-cold PBS and lysed with an immunoprecipitation buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM MgCl 2 , 10 mM NaF, 5% glycerol, 1% Nonidet P-40, 1 mM dithiothreitol, 1 mM EGTA, 1 M microcystin-LR, and mixture proteinase inhibitor). Cell lysates were centrifuged at 10,000 rpm for 10 min to remove insoluble debris. Supernatants were transferred to fresh tubes, precleared with protein G-Sepharose (Invitrogen) beads for 15 min, and incubated with an anti-ROCK antibody (Santa Cruz Biotechnology) for 40 min. The antibody-ROCK complexes were pulled down with protein G-Sepharose beads (20 min). Beads were washed three times in immunoprecipitation buffer and incubated with the kinase reaction mixture (final volume ϭ 25 l) that included 0.25 g of rMYPT1 and 200 g of ATP, fol-lowing the protocol indicated by the manufacturer. Samples were subjected to Western blot, and rMYPT1 phosphorylation was determined using a rabbit anti-phospho-Thr 850 MYPT1 antibody.
Enzyme-linked Immunosorbent Assay-TNF␣ levels were determined by ELISA (Pepro Tech Inc.), essentially as described previously (11). Briefly, 100 l of conditioned medium (CM) were added into each well and incubated overnight at 4°C. Subsequently, 100 l of biotin-labeled anti-TNF␣ antibody (0.25 g/ml) was added for 2 h at room temperature. Bound antibody was detected by incubation with peroxidaselabeled avidine and 2,2Ј-azino-bis(3-ethylbenzothiazoline-6sulfonic acid) from Sigma, and absorbance was measured at 450 nm. Nonspecific binding was blocked with 1% bovine serum albumin in PBS.

PMA Induces the Activation of Rho and ROCK in LNCaP
Prostate Cancer Cells-Phorbol esters induce a strong apoptotic response in LNCaP cells. This effect is mediated primarily by PKC␦, which upon activation stimulates the release of death factors and the extrinsic apoptotic cascade (11). Since PKC isozymes have been shown to regulate Rho GTPases, and Rho has been implicated in apoptosis triggered by diverse stimuli, we speculated that this small G-protein could be involved in phorbol ester-induced apoptosis in prostate cancer cells. To address this issue, we first examined if PMA could promote the activation of RhoA in LNCaP cells. Using a pull-down assay, we found that treatment with PMA (100 nM) caused a robust and sustained elevation in Rho-GTP levels in LNCaP cells, which persisted for at least 3 h poststimulation (Fig. 1A). RhoA activation by PMA was blocked by pretreatment of LNCaP cells with the pan-PKC inhibitor GF 109203X (Fig. 1B), suggesting that it is mediated by PKC isozymes and not other phorbol ester receptors unrelated to PKC (2).
Since ROCK is a main Rho effector, we determined the effect of PMA on ROCK activation in LNCaP cells. As a first approach, we examined the phosphorylation of the ROCK substrate MYPT1, using a specific phosphoantibody against Thr 850 -MYPT1, a well established ROCK phosphorylation site (27). As shown in Fig. 1C (left), PMA induces a marked MYPT1 phosphorylation, which peaks at 15 min and is sustained for at least 1 h. ROCK activation can also be observed in C4-2 and CRW22 androgen-dependent prostate cancer cells (Fig. 1C, right).
Next, we examined ROCK activation using an immunoprecipitation assay. Endogenous ROCK was pulled down from LNCaP cells treated with PMA (100 nM, 0 -15 min), and ROCK activity was determined using an in vitro kinase assay using as a substrate the recombinant fragment of MYPT1 that comprises the ROCK phosphorylation site (amino acids 654 -880). Phosphorylation of the peptide was determined by Western blot using the anti-Thr 850 -MYPT1 antibody. Consistent with the results in Fig. 1C, ROCK kinase activity was enhanced in immunoprecipitates of PMA-treated LNCaP cells (Fig. 1D). Activation of ROCK by PMA was blocked by pretreatment of cells with the ROCK inhibitor Y-27632 (Fig. 1E), as expected, as well as by the pan-PKC inhibitor GF 109203X (Fig. 1F).
Rho GTPases are best known for their roles in the regulation of cytoskeleton rearrangement and the generation of the contractile force necessary for the formation of stress fibers (13). We first examined if PKC activation causes cytoskeleton reorganization in LNCaP cells. As shown in Fig. 1G, PMA promotes the formation of stress fibers, and this effect was sensitive to the PKC inhibitor GF 109203X. Moreover, the ROCK inhibitor Y-27632 blocked the formation of stress fibers in response to PMA. Taken together, these results indicate that PMA is a strong activator of Rho and its downstream effector ROCK in LNCaP prostate cancer cells.
Rho and ROCK Are Implicated in PMA-induced Apoptosis-To determine whether Rho plays a role in PMA-induced apo-ptosis in LNCaP cells, we expressed a dominant negative RhoA mutant (N19-RhoA) by adenoviral means. As a control, we used a GFP AdV. Expression of N19-RhoA reduced the apoptotic effect of PMA by ϳ40% ( Fig. 2A). Next, to determine if ROCK is implicated in PMA-induced apoptosis, we used the ROCK inhibitor Y-27632. As shown in Fig. 2B, the ROCK inhibitor significantly blocked PMA-induced apoptosis in LNCaP cells, suggesting a role for the Rho-ROCK pathway in phorbol esterinduced apoptosis. Y-27632 also blocked the apoptotic effect of PMA in C4-2 and CRW22 cells (supplemental Fig. S1).
PKC␦ Mediates PMA-induced Activation of Rho and Stress Fiber Formation-Previous studies from our laboratory demonstrated that PMA-induced apoptosis is mediated by the acti- vation of PKC␦. Inhibition of PKC␦ by either pharmacological means or expression of a dominant negative PKC␦ mutant reduces PMA-induced apoptosis in LNCaP cells (7,11,28). Likewise, we reported that PKC␦ depletion from LNCaP cells using RNAi markedly impairs the apoptotic response of the phorbol ester (11).
Given the involvement of PKC␦ as a mediator of phorbol ester-induced apoptosis (7, 11, 28), we decided to evaluate if this PKC is implicated in Rho activation by PMA. RhoA-GTP levels were determined in LNCaP cells subject to PKC␦ RNAi, using two different RNAi duplexes (PKC␦ RNAi-1 and PKC␦ RNAi-2). PKC␦ depletion was Ͼ75% with either RNAi duplex (Fig. 2C, left). Fig. 2C (right) shows that the activation of RhoA by PMA was greatly reduced in PKC␦-depleted LNCaP cells. We next assessed the effect of PKC␦ knockdown on ROCK activation. As shown in Fig. 2D, PMAinduced activation of ROCK in LNCaP cells was impaired as a consequence of PKC␦ RNAi depletion. To further assess the involvement of PKC␦ in Rho function, we examined stress fiber formation in response to PMA in PKC␦-depleted cells. Fig.  2E shows that LNCaP cells subjected to PKC␦ RNAi failed to assemble stress fibers in response to the PKC activator. These observations indicate that activation of PKC␦ in LNCaP cells promotes the activation of RhoA and RhoAdependent responses via ROCK.
Cytoskeleton-dependent JNK Activation Is Required for PMA-induced Apoptosis-Our previous studies established that the apoptotic effect of phorbol esters is caused by the autocrine release of death factors, primarily TNF␣ and TRAIL, an effect mediated by PKC␦. CM collected from PMA-treated LNCaP cells has the ability to trigger an apoptotic response via the extrinsic cascade, which becomes activated in response to the death factors released to the CM. Although we found that inhibition of JNK does not affect the release of death factors (11), SP600126 blocked the apoptotic effect of CM from PMA-treated cells, suggesting a key role for the JNK pathway in the effect of the autocrine factors. CM collected from PMAtreated LNCaP cells causes a strong and sustained JNK activation when added to a previously untreated culture of LNCaP cells (11). Fig. 3A shows that the JNK inhibitor SP600126 caused a significant impairment of the PMA apoptotic response. The effect was observed when SP600126 was left in the medium after the PMA treatment but not if only added for short times after PMA treatment (data not shown) (28). These results are consistent with the involvement of JNK as a mediator of the effect of death factors but not their release (11). It has been demonstrated in various cellular models that Rho and ROCK mediate JNK activation in response to stimuli (29 -31). Therefore, we examined if ROCK is involved in JNK activation by PMA in LNCaP cells. As shown in Fig. 3B, the ROCK inhibitor Y-27632 markedly reduced JNK activation.
ROCK plays a major role in the assembly of actin stress fibers, which is required for membrane blebbing and nuclear disintegration and has been shown in some cases to be required for the execution of apoptosis (13)(14)(15)(16)(17)(18)(19). These effects are mediated by ROCK phosphorylation of myosin light chain and activation of its ATPase activity, thus increasing the actin-myosin force and cell contractility (15,18). We therefore speculated that the cytoskeleton is implicated in apoptosis triggered by PKC␦ activation. To address this issue, we used the actin-destabilizing agent cytochalasin B and the myosin II inhibitor blebbistatin. As shown in Fig. 3C, although cytochalasin B or blebbistatin did not cause any measurable apoptotic response by themselves, they markedly inhibit PMA-induced apoptosis in LNCaP cells. This suggests that cytoskeleton rearrangement in response to PMA activation is required for the apoptotic effect of the phorbol ester in LNCaP prostate cancer cells. Furthermore, the cytoskeleton inhibitors cytochalasin B and blebbistatin also blocked JNK activation by PMA (Fig. 3D). Inhibition of cytoskeleton reorganization did not impair PKC␦ activation, as determined by the lack of effect of blebbistatin on PKC␦ trans-location (supplemental Fig. S2). Taken together, these results indicate that ROCK regulates apoptosis through the activation of the JNK pathway through a cytoskeleton-dependent mechanism.
ROCK-mediated p21 Cip1 Up-regulation Is Required for Apoptosis-It is noteworthy that in many cell types, PKC activation leads to the induction of the cell cycle inhibitor p21 Cip1 , which plays a central role in PMA-induced cell cycle arrest and senescence. This induction could be either p53-dependent or p53-independent, depending upon the cell type and stimuli (4,32,33). Studies have shown that p21 Cip1 up-regulation preceded apoptosis induced by PMA in prostate cancer cells. To determine whether a causal relationship between p21 Cip1 and PMAinduced apoptosis exists, we knocked down p21 Cip1 using two different RNAi duplexes (p21 RNAi-a and p21 RNAi-b). A significant depletion in p21 Cip1 levels was observed by delivery of either RNAi duplex into LNCaP cells (Fig. 4A). In LNCaP cells subjected to p21 Cip1 RNAi, the induction of this cell cycle inhibitor was significantly attenuated, although full inhibition of p21 Cip1 up-regulation could not be achieved (ϳ70 -80% depletion). Knockdown of p21 Cip1 led to a reduced PMA apoptotic response (Fig. 4B). The effect was partial, which may suggest a partial requirement for p21 Cip1 or otherwise relate to the residual p21 Cip1 induction by the phorbol ester. The requirement of p21 Cip1 for phorbol ester-induced apoptosis prompted us to investigate whether this effect was ROCK-mediated. We therefore examined the effect of Y-27632 on p21 Cip1 induction. Fig. 4C shows that PMA treatment caused a time-dependent induction of p21 Cip1 in LNCaP cells. This induction was markedly reduced in the presence of the ROCK inhibitor (see also Fig.  4E). To determine whether p21 Cip1 up-regulation by PMA is dependent upon the cytoskeleton rearrangement, we examined the effect of cytochalasin B and blebbistatin. Interestingly, both cytoskeleton inhibitors prevented p21 Cip1 upregulation by the phorbol ester (Fig. 4D). The induction of p21 Cip1 by PMA was not affected by the JNK inhibitor SP600125 (Fig. 4E), suggesting that JNK activation and p21 Cip1 induction downstream of Rho/ROCK are implicated in parallel mechanisms rather than in a linear pathway.
ROCK Regulates p21 Cip1 at a Transcriptional Level-Expression of p21 Cip1 is controlled both by transcriptional and posttranscriptional mechanisms, including in response to PMA (4,32,33). Previous studies from our laboratory established a key role for PKC␦ in p21 Cip1 induction. PMA caused a marked ele- vation in p21 Cip1 mRNA levels, as determined by real time PCR (4.0 Ϯ 0.9-fold increase at 3 h; 7.9 Ϯ 0.6-fold increase at 6 h; n ϭ 3). To determine whether p21 Cip1 induction by PMA has a transcriptional component, we carried out a gene reporter assay using a p21 Cip1 promoter luciferase reporter. LNCaP cells were transfected with the p21 Cip1 reporter vector p21P (25), which encodes a fragment comprising bp Ϫ1 to Ϫ2400 in the human p21 Cip1 promoter fused to firefly luciferase. A Renilla luciferase vector was co-transfected for normalization of transfection efficiency. As shown in Fig. 5A, PMA treatment led to a marked increase in luciferase activity. Treatment with either Y-27632 or cytochalasin inhibited luciferase activity to a great extent.
Transcriptional activation of p21 Cip1 is mediated through multiple mechanisms. Among the most relevant transcription factors, p53 and Sp1 have been widely implicated in transcriptional control of the p21 Cip1 promoter (32,34,35). We used two deletion mutants of the human p21 Cip1 promoter. The first mutant (p21PSma) has a deletion in bp Ϫ114 to Ϫ2400, which comprises the p53-responsive element in the promoter. The second mutant (p21PSma⌬1) has the four Sp1 elements deleted (deletion in bp Ϫ49 to Ϫ114). Notably, although deletion of the large fragment that includes the p53-binding site caused only a minimal effect, deletion of the Sp1-responsive elements abrogated the induction of luciferase activity by PMA (Fig. 5B). To demonstrate the involvement of Sp1 in p21 Cip1 induction, we used RNAi to knock down endogenous Sp1. We used two RNAi sequences: Sp1 RNAi-1, which depleted endogenous Sp1 by Ͼ80%, and Sp1 RNAi-2, which was less efficient (Fig. 5C, left). We found that Sp1 RNAi-1 significantly blocked p21 Cip1 up-regulation, whereas Sp1 RNAi-2 had only a marginal effect (Fig. 5C,  right). These results suggest that the induction of p21 Cip1 by PMA is mediated by Sp1-dependent mechanisms and independent of p53, despite the fact that LNCaP cells have functional p53.

ROCK Regulates TNF␣ Secretion in a JNK-independent
Manner-Previous studies from our laboratory demonstrated that PMA-induced apoptosis of LNCaP cells is triggered by the autocrine secretion of death factors via PKC␦, with the subse-  OCTOBER 23, 2009 • VOLUME 284 • NUMBER 43 quent activation of the extrinsic apoptotic cascade. Among the death factors, TNF␣ is the most relevant one. JNK was identified as a mediator of the TNF␣ effect (11). To examine whether ROCK regulates TNF␣ secretion via JNK, we compared TNF␣ levels in conditioned medium from LNCaP cells treated with PMA in the absence or presence of the ROCK inhibitor Y-27632 or the JNK inhibitor SP600125. As shown in Fig. 6A, ROCK inhibitor markedly inhibited TNF␣ secretion, whereas the JNK inhibitor had no effect. These results suggest that ROCK regulates TNF␣ secretion in a JNKindependent manner.

DISCUSSION
Despite the well established role for PKC in mitogenesis and tumor promotion, considerable evidence supports the involvement of discrete PKC isozymes in growth inhibition and apoptosis. Phorbol esters exert profound effects on the machinery that controls the cell cycle and modulate in either positive or negative manners multiple signaling pathways implicated in proliferation, survival, and apoptosis. In that regard, studies by our laboratory and others have established a prominent role for PKC␦, a member of the novel PKC family, as a mediator of growth inhibition and apoptosis in response to phorbol esters (4 -8, 11, 28). Androgen-dependent prostate cancer cells, such as LNCaP and CWR22-Rv1 cells, undergo apoptosis in response to phorbol esters via activation of PKC␦ (7-9). The signaling components implicated in PKC␦-driven apoptosis are only partially understood. Previous studies from our laboratory established key roles for the JNK cascade as a mediator of phorbol ester-induced apoptosis in LNCaP cells (11,28). A more detailed analysis established that apoptosis is triggered by the autocrine secretion of death factors from prostate cancer cells via PKC␦, including TNF␣ and TRAIL, with the subsequent activation of the extrinsic apoptotic cascade. Interfering with death receptor signaling either by inhibition/depletion of TNF␣/TRAIL receptors, RNAi depletion of the adaptor FADD or caspase-8, or pharmacological inhibition of p38 MAPK and JNK pathways in all cases impairs phorbol ester-induced apoptosis (11). A distinctive aspect of the present study is the identification of the Rho-ROCK-p21 cip1 pathway as an effector of PMA-induced apoptosis, which provides an additional layer of complexity and argues for the involvement of an intricate network of signaling cascades in this effect.
Our results provide evidence that PMA induces RhoA and ROCK activation as well as the formation of stress fibers in LNCaP prostate cancer cells via ROCK. The activation of the Rho pathway and cytoskeletal organization by PMA is mediated by PKC␦. Both the pan-PKC inhibitor GF 109203X and PKC␦ RNAi depletion block RhoA activation and stress fiber formation. PKC isozymes have been implicated in Rho activation in several models, including cardiomyocytes (36), myoblasts (37), and endothelial cells (38). Although the mechanistic basis of PKC␦ modulation of Rho/ROCK function in LNCaP still remains to be elucidated, it has been shown that phorbol esters can activate Rho GTPases through different mechanisms that involve either Rho-GEF activation or Rho-GDI inactivation (22,23). Rho-GEFs are a very diverse family, with Ͼ70 members being identified in humans (39). Studies have shown that PMAinduced apoptosis in erythroblastic D2 cells involves the RhoA-GEF GEF-H1. Depletion of GEF-H1 leads to a significant reduction of RhoA activity and prevents cells from undergoing PMA-induced apoptosis via the Rho-ROCK pathway (22). Studies in vascular smooth muscle cells revealed that Rho-dependent migration and JNK activation induced by angiotensin II were inhibited by the PKC␦ inhibitor rottlerin or a dominant negative PKC␦ mutant. PKC␦-dependent Rho/ROCK activation in these cells possibly involves PDZ-RhoGEF (40). There is also evidence that a PKC phosphorylation site in Rho-GDI modulates its association/dissociation from RhoA by changing the affinity for the small G-protein. Upon phosphorylation in Ser 96 , Rho-GDI dissociates from RhoA, thereby enabling RhoA activation in response to stimuli (23). Conceivably, PKC␦ may also regulate the activity of Rho-GAPs and control the rate of GTP hydrolysis from the GTPase. For example, phorbol esters stimulate the phosphorylation of the Rho-GAP DLC1, leading to its relocalization and suppression of GAP activity. Several studies have also reported a physical association between PKCs and RhoA (24,41). Although there is very limited information regarding the expression and regulation of Rho-GEFs in prostate cancer cells, it is conceivable that mechanisms such as those described above and/or alternative yet unknown mechanisms may operate in prostate cancer cells, which requires further investigation, since they should provide relevant information on how PKC␦ modulates RhoA in the context of apoptotic responses. Candidate Rho GEFs include Vav3, which has been implicated in prostate cancer progression (42), as well as LARG and PDZ-Rho-GEF, which mediate downstream Rho signaling in prostate cancer models (43,44).
ROCK is a major regulator of cytoskeleton reorganization in response to stimuli. Studies have implicated the Rho/ROCK pathway in apoptosis in various cell models, including prostate cancer cells (19,45), and established links between cytoskeleton remodeling and apoptosis programming or execution (17)(18)(19). ROCK has been implicated in myosin light chain phosphorylation and membrane blebbing as well as in Golgi fragmentation during apoptosis (47,48). It is also known that ROCK modulates the extrinsic apoptotic cascade, which mediates PMA-induced apoptosis in LNCaP prostate cancer cells via the autocrine release of death factors and activation of caspase-8 (11). Studies in a paradigm of PMA-induced apoptosis in erythroblastic TF-1 cells implicated ROCK in the formation of DISC complex (19). Interestingly, we also found that ROCK inhibition prevents the release of death factors from LNCaP cells in response to PMA (data not shown). Since this release of death factors is entirely dependent on PKC␦ (11), we speculate that ROCK is also downstream of PKC␦ in the regulation of the autocrine response. Studies have shown that the cytoskeleton is involved in the extrinsic apoptotic cascade, since it controls DISC formation and modulates the localization of key components of the cascade, including death receptors (49 -51). The Rho pathway also controls transcriptional mechanisms via cytoskeleton-dependent and -independent processes (29,52,53), and a ROCK-dependent transcriptome has been recently defined, which includes genes involved in cytoskeletal reorganization (54). Actin polymerization modulates the activity of RNA polymerases and has been also implicated in chromatin remodeling as well, since it serves as a platform for the cotranscriptional recruitment and/or tethering of histone-modifying enzymes via its association with pre-mRNPs (55). Interestingly, in PMA-treated proapoptotic D2 cells, two nuclear restricted pre-mRNA binding proteins, heterogeneous nuclear ribonucleoprotein C1 and C2, are translocated to the cytosolic compartment in a ROCK-dependent manner (56). Notably, analysis of PKC-regulated genes in LNCaP cells revealed a number of Rho-GEFs that become up-regulated in response to PMA, including ARHGEF2, ARHGEF10, and ARHGEF12, as well as Rho-GAPs that become down-regulated, such as ARHGAP11a and ARHGAP19. 4 The relative contributions of these Rho modulators in prostate cancer cells still need to be determined.
Our studies also implicate the cell cycle inhibitor p21 Cip1 as a mediator of PMA-induced apoptosis in prostate cancer cells. It is well established that in many cell types, PMA stimulation leads to the induction of p21 Cip1 , and this induction plays a central role in PMA-induced growth inhibition (4,5). Previous studies from our laboratory in lung cancer cells showed that PKC␦ controls p21 Cip1 induction at a transcriptional level, whereas other PKCs are less important for controlling the expression of this cell cycle inhibitor (4). Early studies established that p21 Cip1 induction and retinoblastoma dephosphorylation preceded PMA-induced apoptosis in LNCaP cells (10). We found that impeding p21 Cip1 induction using RNAi attenuates cell death in response to the phorbol ester. Recently, studies in prostate and colon cancer cell lines showed that p21 Cip1 up-regulation is essential for TRAIL sensitization by proteasome inhibitors (57) and HDAC inhibitors (58), highlighting the relevance of this cell cycle inhibitor in sensitizing for cell death induced by cytokines, as suggested by our studies. We also determined that p21 Cip1 up-regulation by PMA is ROCK-dependent. The regulation of p21 Cip1 by Rho family members is controversial, since Rho was found to either suppress or induce the expression of cell cycle inhibitors, depending on the cell type (59 -61). In contrast to our studies in prostate cancer cells, PMA-treated D2 erythromyeloblasts, the proapoptotic population has impaired p21 Cip1 up-regulation resulting from ROCK activation, suggesting major cell type differences in the control of the pathway. JNK has been also linked to p21 Cip1 expression (62). However, although in LNCaP cells JNK activation is ROCK-dependent, a JNK inhibitor does not abrogate p21 Cip1 induction, suggesting the involvement of parallel pathway(s). Although p21 Cip1 up-regulation in some cell types is dependent upon the activation of the MEK-ERK pathway (60), Y-27632 did not affect ERK phosphorylation by PMA in LNCaP cells (data not shown), suggesting that in our model, ROCK regulation of p21 Cip1 is ERK-independent. Indeed, our previous studies showed that a MEK inhibitor enhances PMA-induced apoptosis (28). A p38 inhibitor also failed to impair p21 Cip1 induction by PMA in LNCaP cells (data not shown). Although MAPK cascades exert transcriptional effects via Sp1 transcription factors, deletion of Sp1 sites in the human p21 Cip1 promoter abrogates the induction by PMA. As PMA exerts pleiotropic effects on signaling cascades, it is conceivable that MAPK-independent pathways modulate p21 Cip1 transcriptional activation, such as direct phosphorylation of Sp1 by protein kinase G, cyclin-dependent kinases, or other kinases (46,63,64).
In summary, our studies showed that RhoA/ROCK and the cytoskeleton are implicated in PMA-induced apoptosis in prostate cancer cells. A key role for the cell cycle inhibitor p21 Cip1 has also been identified in the context of Rho signaling. A model summarizing our findings is depicted in Fig. 6B. A considerable amount of effort has been devoted to developing PKC modulators with anti-cancer therapeutic value, although selective agonists for PKC␦ as antimitogenic/proapoptotic agents have yet to be developed. The elucidation of the PKC␦ downstream effectors is highly relevant to understanding the basis of cell growth inhibition and apoptosis mediated by this kinase. The identification of Rho/ROCK as a mediator of PKC␦-driven apoptosis may provide a novel platform for the design of PKC modulators for cancer treatment.