P21-Activated Kinase 4 Promotes Prostate Cancer Progression Through CREB

ORIGINAL ARTICLE p21-Activated kinase 4 promotes prostate cancer progression through CREB

M-H Park1,6, H-S Lee2,6, C-S Lee1,STYou1, D-J Kim1, B-H Park2, MJ Kang3, WD Heo4, E-Y Shin1, MA Schwartz5 and E-G Kim1

Prostate cancer is initially androgen-dependent but, over time, usually develops hormone- and chemo-resistance. The present study investigated a role for p21-activated kinase 4 (PAK4) in prostate cancer progression. PAK4 activation was markedly inhibited by H89, a speciﬁc protein kinase A (PKA) inhibitor, and PAK4 was activated by the elevation of cAMP. The catalytic subunit of PKA interacted with the regulatory domain of PAK4, and directly phosphorylated PAK4 at serine 474 (S474). Catalytically active PAK4 enhanced the transcriptional activity of CREB independent of S133 phosphorylation. Stable knockdown of PAK4 in PC-3 and DU145 prostate cancer cells inhibited tumor formation in nude mice. Decreased tumorigenicity correlated with decreased expression of CREB and its targets, including Bcl-2 and cyclin A1. Additionally, in androgen-dependent LNCap-FGC cells, PAK4 regulated cAMP- induced neuroendocrine differentiation, which is known to promote tumor progression. Finally, PAK4 enhanced survival and decreased apoptosis following chemotherapy. These results suggested that PAK4 regulates progression toward hormone- and chemo-resistance in prostate cancer, and this study identiﬁed both a novel activation mechanism and potential downstream effector pathways. Therefore, PAK4 may be a promising therapeutic target in prostate cancer.

Oncogene (2013) 32, 2475–2482; doi:10.1038/onc.2012.255; published online 18 June 2012 Keywords: prostate cancer; p21-activated kinase 4; protein kinase A; CREB; neuroendocrine differentiation; chemoresistance

INTRODUCTION cancer cell migration in response to hepatocyte growth factor 14 Prostate cancer is the most common cancer and a leading cause (HGF). However, the mechanism of PAK4 activation and its of cancer deaths in men in the developed world.1 Hormonal downstream effectors in prostate cancer are largely unknown. therapy for advanced prostate cancer starts with medical or Activation of group II PAKs is believed to involve autopho- surgical castration, but androgen-independence eventually sphorylation at serine 474 (S474) for PAK4, S602 for PAK5 and 15 S474E develops in most patients.2,3 Androgen-independent metastatic S560 for PAK6. Thus, PAK4 may represent an active phos- 12 prostate cancers typically respond poorly to chemotherapy, so phomimetic form. PAK4 can also be constitutively activated by improved therapies are badly needed. Developing new treatments the mutation of S445 to N445 in the catalytic domain, which will require a deeper molecular understanding of tumor progression. enhances the catalytic activity by stabilizing the PAK4-substrate 8 p21-Activated kinases (PAKs) comprise a family of serine/ complex. However, it is also possible that an upstream kinase(s) threonine kinases that are classiﬁed into the following two groups may phosphorylate PAK4 at S474 to regulate its activity. that have distinct structural and functional features: group I (PAK1- In the present study, we sought to elucidate the mechanism of 3) and group II (PAK4-6).4 Group I PAKs contribute to a wide range PAK4 activation and the role of PAK4 in the progression of prostate of cellular processes including actin cytoskeleton reorganization cancer. These results identiﬁed an unexpected relationship with and tumorigenesis.5 Group I PAKs are present as autoinhibited, protein kinase A (PKA), which was found to function as both an inactive dimers mediated by the binding of the kinase domain to upstream activator and a partner in downstream signaling. an inhibitor segment in the regulatory domain. When this inhibition is relieved, for instance, by Cdc42 binding, they undergo dissociation into active monomers.6,7 By contrast, RESULTS group II PAKs are monomeric and have some basal activity, PKA functions upstream of PAK4 because they contain no known autoinhibitory segment.4 To understand the mechanism of PAK4 activation in prostate The group II member, PAK4, also regulates the actin cytoske- cancer, we screened prostate cancer cell lines in which leton and contributes to tumorigenesis.8–11 PAK4 is overexpressed endogenous PAK4 was activated. To this end, we performed a in various human cancer cell lines such as the lung, ovary, colon kinase assay and immunoblotting. PC-3, DU145 and LNCap-FGC and breast cells.12 In ovarian cancer cells, PAK4 regulates cell lines showed similar total PAK4 levels (Figure 1a). However, proliferation, migration and invasion and PAK4 correlates with PAK4 kinase activity and the levels of S474-phosphorylated PAK4 poor prognosis.13 PAK4 has also been shown to promote prostate (pPAK4S474) were elevated in PC-3 and DU145 cells compared with

1Department of Biochemistry and Medical Research Center, Chungbuk National University College of Medicine, Cheongju, Korea; 2Department of Biochemistry, Chonbuk National University Medical School, Jeonju, Jeonbuk, Korea; 3Department of Pathology, Chonbuk National University Medical School, Jeonju, Jeonbuk, Korea; 4Department of Biological Sciences, KAIST, Daejeon, Korea and 5Departments of Medicine and Cell Biology, Yale Cardiovascular Research Center, Yale University School of Medicine, New Haven, CT, USA. Correspondence: Professor E-G Kim, Medical Research Center and Department of Biochemistry, Chungbuk National University College of Medicine, Cheongju, Chungbuk 361-763, Korea. or Professor E-Y Shin, Medical Research Center and Department of Biochemistry, Chungbuk National University College of Medicine, Cheongju, Chungbuk 361-763, Korea E-mail: [email protected] or [email protected] 6These authors contributed equally to this work. Received 29 November 2011; revised 21 March 2012; accepted 20 April 2012; published online 18 June 2012 Role for PAK4 in prostate cancer progression M-H Park et al 2476

Figure 1. Regulation of PAK4 by PKA in prostate cancer cells. (a) PAK4 kinase assay was performed (top) and immunoprecipitates were S474 immunoblotted for pPAK4 or total PAK4 (bottom). (b) DU145 cells were treated with LY294002 (20 mM), PD98059 (50 mM), PP2 (10 mM), GF109203X (2.5 mM) and H89 (20 mM) for 1 h (top) or with the indicated concentrations of H89 for 1 h (bottom). Lysates were analyzed for pPAK4S474 and total PAK4. (c) Lysates were immunoblotted for pPKA-CT197 and total PKA-C. (d) DU145 cells were incubated with the indicated concentrations of H89 for 1 h and then stimulated with 1 mM cAMP for 30 min. Lysates were immunoblotted as described above. The pPAK4S474 bands were quantiﬁed by densitometric analysis. GAPDH was used as a loading control (a–c).

LNCap-FGC cells (Figure 1a). PAK4 kinase activity correlated with association was markedly reduced upon pre-incubation with H89 the levels of pPAK4S474, as reported previously.12 Therefore, we (Figure 2b), which suggested that this association may require PKA monitored PAK4 activation by immunoblotting for pPAK4S474 in activity. Conversely, the PAK4 regulatory domain (PAK4-R) but subsequent experiments. Screening DU145 cells with various not the PAK4 catalytic domain (PAK4-C) interacted with PKA-C kinase inhibitors showed that the phosphorylation of PAK4 at (Figure 2c). S474 was most strongly decreased by H89, a PKA inhibitor To visualize this association in cells, we employed the ECLIPSE (Figure 1b, top). We also observed some inhibition by the system. In the ECLIPSE system, PKA-R or PKA-C in an AmCyan- extracellular signal-regulated kinase blocker, PD98059, but the tagged form is recruited to endosomes in a rapamycin-dependent effect was weaker and less consistent and, therefore, were not manner, and the recruitment of YFP-tagged PAK4 to this pursued further. site is assayed.16 We observed the rapamycin-dependent On the basis of the H89 result, we next examined PKA as a formation of ‘ecliptical rings’ in cells expressing PAK4-R and possible upstream regulator of PAK4. As illustrated in Figure 1b PKA-C but not in cells expressing PAK4-R and PKA-R (bottom), PAK4 phosphorylation was inhibited by H89 in a dose- (Supplementary Figure S2). Thus, these data suggested that dependent manner. To investigate whether PKA was activated in PKA-C and PAK4-R specifically associate in cells. To determine these cells, we examined its phosphorylation on T197, which is a whether this interaction was direct, we examined the interaction key regulatory site. Consistent with PAK4 phosphorylation, PC-3 using purified recombinant proteins. Full-length PAK4 and and DU145 cells had a phosphorylated catalytic subunit of PKA PAK4-R co-precipitated with PKA-C (Supplementary Figure S3), (PKA-C) at T197 compared with LNCap-FGC (Figure 1c). Therefore, thus supporting a direct interaction. Finally, to understand PKA and PAK4 activities were correlated. To further test whether PKA the physiological relevance of this association, we examined the was an upstream kinase for PAK4, we activated PKA by dibutyryl- endogenous association and its dependence on stimulation cAMP (cAMP) or depleted PKA-C by short interfering RNA (siRNA). with a physiological agonist such as dihydrotestosterone (DHT), Treatment with cAMP increased the levels of pPAK4S474,andthis which is an androgen analog. As shown in Figure 2d, PAK4 effect was blocked by H89 in a dose-dependent manner (Figure 1d, specifically bound PKA-C in DU145 and PC-3 cells. Because lanes 2–5). A similar result was obtained upon stimulation by androgen has been reported to stimulate PKA activation in forskolin (Fsk) (Figure 4c and S8A). Conversely, depletion of PKA-C LNCap-FGC cells,17 the DHT effect on the PKA–PAK4 association with two different siRNAs markedly reduced PAK4 phosphorylation was tested in these cells. Both PKA and PAK4 were activated (Supplementary Figure S1). Taken together, these data demonstrated 15 min after DHT treatment and remained active until 2 h that PKA is upstream of PAK4. (Supplementary Figure S4). Interestingly, pre-incubation with H89 completely blocked the DHT-stimulated activation of PAK4, which indicated that PKA functions upstream of PAK4 PKA binds and phosphorylates PAK4 at serine 474 in the androgen signaling pathway. Initially, we performed a We next investigated whether the PKA and PAK4 kinases binding test without any transfection, but this was unsuccessful interacted. To this end, 293T cells were transfected with various because of the low level of endogenous PKA-C in LNCap-FGC GFP- and myc-tagged constructs of PKA and PAK4, and their cells (Figure 1c). When PKA-C was overexpressed, we observed association was monitored by co-immunoprecipitation. The a DHT-dependent increase in the association between catalytic subunit of PKA (PKA-C) but not the regulatory subunit PKA-C and PAK4, which was markedly inhibited by H89 (PKA-R), was present in PAK4 immunoprecipitates (Figure 2a). This (Figure 2e).

Oncogene (2013) 2475 – 2482 & 2013 Macmillan Publishers Limited Role for PAK4 in prostate cancer progression M-H Park et al 2477

Figure 2. PKA binds and phosphorylates PAK4. (a–c), 293T cells were transiently co-transfected with plasmids encoding the proteins as indicated. After 24 h, cells were incubated with 1 mM cAMP for 30 min (a). H89 was included for 1 h before lysis (b). Lysates were immunoprecipitated with an anti-Myc antibody and were analyzed by immunoblotting with an anti-GFP or anti-Myc antibody. (d) Cells were immunoprecipitated with the indicated antibodies and were probed for PKA-C and PAK4. (e) LNCap-FGC cells were transfected with the plasmid encoding GFP-PKA-C. After 24 h, cells were stimulated with DHT (1 mM) for 30 min in the absence or presence of H89 and then processed as described above. (f) Potential phosphorylation sites on PAK4 by PKA (arrowheads). (g)Anin vitro kinase assay was performed using active PKA-C (50 ng) and kinase-inactive GST-PAK4K350M, which contains mutations as indicated. Phosphorylation was detected by autoradiography (top). The Ponceau stain demonstrates equal loading (bottom). (h) 293T cells expressing Myc-PAK4-WT were stimulated with increasing doses of Fsk for 30 min (lanes 1–5) or were pre-incubated with 20 mM H89 (lane 5) for 1 h before stimulation. Lysates were immunoprecipitated and immunoblotted using an anti-pPAK4S474 (top), anti-RRXpS/pT (middle) or anti-Myc antibody (bottom).

The binding data raised the possibility that PKA may phosphorylation was almost completely blocked by pre-incuba- phosphorylate PAK4. Indeed, PAK4 contains the following tion with H89 (lane 5). Taken together, these data showed that two consensus sequences that correspond to the preferred PKA phosphorylates PAK4 at S474, which is a site implicated in PKA substrates: RRXS/T (in which X stands for any residue), PAK4 activation.12 S104 and S474 (Figure 2f). These serines were mutated to alanines in catalytically inactive PAK4K350M and the mutants were PAK4 contributes to prostate tumorigenesis incubated with active PKA. We observed a strong phosphorylation To test the role of PAK4 in tumorigenesis, we established stable of PAK4 by PKA that was completely blocked by the S474A PAK4 knockdown PC-3 and DU145 cell lines (PC-3-shPAK4 and mutation (Figure 2g). In contrast, the S104A mutation had DU145-shPAK4) and monitored their tumor formation in athymic no effect. We next examined this phosphorylation in cultured mice. The short hairpin RNAs (shRNAs) led to an almost complete 293T cells. Elevation of cAMP with Fsk triggered phosphorylation loss of PAK4 in both PC-3 and DU145 cell lines, and PAK4 in the of PAK4K350M on S474 (Supplementary Figure S5). As expected, scrambled control lines (PC-3-shScr and DU145-shScr) was this phosphorylation was absent in S474A mutants but not in unaffected (Figure 4c and Supplementary Figure S8). The growth S104A mutants. Furthermore, in cells expressing myc-PAK4, Fsk of PC-3-shPAK4 cells in mice was drastically decreased relative to caused a dose-dependent increase in PAK4 phosphorylation, controls (Figure 3). In DU145 cells, the knockdown of PAK4 led to a which was detected by both an anti-pPAK4S474 antibody and virtually complete blockade of tumor growth (Supplementary an anti-RRXpS/pT antibody that recognizes the phospho-PKA Figure S6). These data suggested that PAK4 is critical in prostate consensus sequences (Figure 2h, lanes 1–4). Moreover, this tumorigenesis.

& 2013 Macmillan Publishers Limited Oncogene (2013) 2475 – 2482 Role for PAK4 in prostate cancer progression M-H Park et al 2478

Figure 3. Reduction of tumorigenicity of PC-3 cells after PAK4 knockdown. (a) Photographs of mice bearing tumors. (b) Growth curves of PC-3-shScr and PC-3-shPAK4 xenografts in athymic mice. Values are the means±s.d. *Po0.05 and **Po0.01 vs PC-3-shPAK4.

PAK4 regulates protein level and transcriptional activity of CREB both the expression of neuron-specific enolase (NSE), a NE PKA activates the transcriptional activity of CREB by phosphorylat- differentiation marker, and the extension of neurite-like ing CREB on serine 133.18 The association of PAK4 with PKA led us processes. NSE levels increased in response to both cAMP and to test whether PAK4 participated in this pathway. The expression active PAK4, which functioned in a dose-dependent manner of PKA-C elevated CREB activity B3-fold over the control vector (Figure 5a, top). Conversely, PAK4 knockdown blocked the cAMP- (Figure 4a). A dominant active form of PAK4S445N also stimulated induced upregulation of NSE (Figure 5a, bottom). Stable knock- CREB activity by nearly the same amount. Contrary to our down of PAK4 induced a similar change in NSE levels (Figure 4c expectation, PAK4S474E did not significantly affect the transcrip- and Supplementary Figure S8a). In accordance with this result, tional activity of CREB, suggesting that this mutant does not NSE staining was decreased in PAK4 knockdown xenograft tumors function as an active phosphomimetic form. However, PAK4S474A (Figure 4d). PAK4 inhibition using a dominant negative construct strongly reduced CREB activity. Therefore, these data suggested also blocked cAMP-stimulated growth of neurite-like processes that PAK4 activity regulates CREB-dependent gene expression. (Figure 5b, left). Quantification confirmed that cAMP approxi- Because the phosphorylation of CREB at S133 is essential for its mately doubled neurite length in cells transfected with PAK4-WT, transcriptional activity, we hypothesized that PAK4 may also and no change occurred in cells expressing dominant negative phosphorylate CREB at this site. However, we could not detect PAK4 (Figure 5b, right). Taken together, these results indicated phosphorylation of CREB even when using active PAK4S445N (data that PAK4 functions downstream of cAMP. Because NE differentia- not shown). Therefore, we considered other molecular mechan- tion correlates with androgen-independence and poor prognosis, isms. Because LNCap-FGC cells showed low PAK4 activation elevated PAK4 activity may explain, in part, the development of (Figure 1a), we overexpressed wild-type PAK4 (PAK4-WT) and the hormone-refractory state in prostate cancer. active PAK4, and we examined the effects of PAK4 overexpression. Overexpression of PAK4-WT and active PAK4 significantly increased the levels of CREB compared with vector control Stable PAK4 knockdown sensitizes prostate cancer cells to chemotherapeutics (Figure 4b, left, and Supplementary Figure S7). PAK4-WT expres- Bcl-2, an anti-apoptotic member of the Bcl family of proteins, is sing cells also displayed high levels of phosphorylation at S474, 21 thus suggesting PAK4 autophosphorylation and activation due to transcriptionally regulated by CREB. Because PAK4 depletion overexpression. Conversely, expression of dominant negative reduced the expression of both CREB and Bcl-2 (Figure 4c and PAK4K350M/S474A strongly downregulated CREB in PC-3 cells Supplementary Figure S8A), we examined the influence of PAK4 (Figure 4b, right). CREB was also downregulated in the stable on cell survival. Basal apoptosis increased from 1 to 6% for PC3- PAK4 knockdown PC-3-shPAK4 cells and in xenograft tumors shScr vs shPAK4 cells, and from 5 to 15.4% for DU145-shScr vs formed from these cells (Figures 4c and d). Accordingly, the shPAK4 cells (Figure 5c). An examination of xenografts showed expression of the CREB target genes, cyclin A1 and Bcl-2, was that PC-3-shPAK4 tumors displayed faint Bcl-2 staining and 24% significantly reduced (Figure 4c). Similar results were obtained terminal deoxynucleotidyl transferase dUTP nick end labeling using the stable PAK4 knockdown DU145-shPAK4 cells (TUNEL)-positive cells per field and that the control PC-3-shScr (Supplementary Figure S8). tumors presented substantially stronger Bcl-2 staining and 2% TUNEL positive cells per field (Figure 5d). We next investigated sensitivity to chemotherapeutic drugs. PAK4 knockdown greatly PAK4 contributes to neuroendocrine differentiation increased sensitivity to docetaxel, paclitaxel and doxorubicin Long-term treatment with cAMP stimulates neuroendocrine (NE) (Figure 5e). The percent apoptosis after treatment with docetaxel, differentiation of LNCap-FGC cells.19,20 Therefore, we investigated paclitaxel and doxorubicin were 13%, 7% and 4%, respectively, for whether PAK4 promoted this process. To this end, we monitored control PC-3-shScr cells, and 29%, 22% and 15%, respectively, for

Oncogene (2013) 2475 – 2482 & 2013 Macmillan Publishers Limited Role for PAK4 in prostate cancer progression M-H Park et al 2479

Figure 4. Regulation of CREB by PAK4. (a) 293T cells were co-transfected with the CRE reporter and Renilla-luciferase together with the plasmids encoding proteins as indicated. After 24 h, CRE reporter activity was assayed. DA, Myc-PAK4S445N; E, Myc-PAK4S474E; A, Myc-PAK4S474A. (b) Left, LNCap-FGC cells were transfected with empty vector or the plasmid encoding Myc-PAK4-WT for 3 days. Lysates were immunoblotted. Right, PC-3 cells were transfected with empty vector or the plasmid encoding dominant negative PAK4 (PAK4-DN, PAK4K350M/S474A) for 3 days and were stimulated with 1 mM cAMP for 30 min before collection. Expression of the indicated protein was analyzed by immunoblotting. (c) PC-3-shScr and PC-3-shPAK4 cells were starved for 24 h and stimulated with 20 mM Fsk for 30 min, or were pre-incubated with 20 mM H89 for 1 h before stimulation. Lysates were immunoblotted for the indicated protein. (d) Xenograft tumors from PC-3-shScr and PC-3-shPAK4 cells were stained for PAK4, CREB, and NSE. The tissue architecture was visualized with H&E staining. Scale bar, 10 mm.

PC-3-shPAK4 cells. DU145 cells were more sensitive to chemotherapy Previously, PAK4S474E has been shown to stimulate the but responded in a similar pattern (Supplementary Figure S9). anchorage-independent growth of NIH3T3 cells,12 thus suggest- Therefore, these data demonstrated that PAK4 inhibition increases the ing that it is constitutively active. In the present study, however, sensitivity of advanced prostate cancer cells to chemotherapy. PAK4S474E was not constitutively active, at least in terms of CREB activation. In contrast, PAK4S474A effectively suppressed CREB activity. The explanation for this discrepancy is currently unknown. DISCUSSION PAK4 activation may require factors other than the addition of a In the present study, we demonstrated that PAK4 can contribute negative charge. Furthermore, the kinase-independent effects of to the progression of prostate cancer by regulating NE differentia- PAKs are well known and may mediate some of these responses.24 tion and cell survival both basally and in response to chemother- Further studies are needed to resolve these issues. apeutic agents. An analysis of the signaling pathway revealed PAK4 may have a role in the nucleus based on its nuclear PAK4 as a major downstream effector of PKA. Active PAK4 localization.22,25 The present study identified CREB as a target of together with PKA enhanced CREB transcriptional activity. The PAK4. CREB is a transcriptional factor that regulates a wide range results suggested that PAK4 increases the expression of CREB, of cellular activities including survival, proliferation and differenti- which is phosphorylated and activated by PKA. Thus, PKA ation. PKA was initially identified as a kinase that regulates the activated CREB-dependent transcription through both direct transcriptional activity of CREB in an S133 phosphorylation- phosphorylation and PAK4-dependent expression. dependent manner.18 Subsequent studies have revealed the PAK4 is activated in response to various agonists, but its direct involvement of other kinases in this phosphorylation.26 However, upstream regulators have remained elusive. In HGF signaling, we could not detect the phosphorylation of CREB by PAK4 in vitro. PAK4 can be activated through phosphoinositide-3-kinase Therefore, it is likely that PAK4 may function in an S133 (PI3K).22 However, PI3K does not seem to mediate PAK4 phosphorylation-independent manner or through the upregu- activation in the prostate cancer cell lines used in this study, lation of CREB. Because CREB represents a promising target for because LY294002 had no effect (Figure 1b). PAK4 may also be cancer therapy and because many small molecules that inhibit activated by associating with the scaffold protein, Gab1, during CREB-mediated gene transcription are under development,27 HGF-induced epithelial cell migration and invasion.23 However, PAK4 inhibitors may prove to be effective tools for targeting this association is independent of PAK4 kinase activity, although it CREB.28 requires Gab1 phosphorylation. The present study identified PKA Development of the hormone-refractory stage is a serious as an upstream regulator of PAK4. challenge to prostate cancer therapy. Accumulating evidence

& 2013 Macmillan Publishers Limited Oncogene (2013) 2475 – 2482 Role for PAK4 in prostate cancer progression M-H Park et al 2480

Figure 5. PAK4 in neuroendocrine differentiation and chemoresistance. (a) Top, LNCap-FGC cells were transfected with the indicated amounts S445N/S474E of Myc-PAK4 plasmid and were analyzed after 3 days. Untransfected cells were stimulated with 1 mM cAMP for 3 days as a positive control (lane 1). Lysates were immunoblotted with anti-NSE or anti-Myc antibody. Bottom, LNCap-FGC cells were untransfected or transfected with scrambled (Scr), or PAK4-speciﬁc siRNA. The cells were stimulated with cAMP for 3 days as indicated. Lysates were immunoblotted for NSE, pPAK4S474 or total PAK4. (b) Left, LNCap-FGC cells were transfected with the plasmid encoding PAK4-WT or PAK4-DN for 3 days in the absence or presence of cAMP. Cells were stained with an anti-Myc antibody (green) and for F-actin (red). Scale bar, 20 mm. Right, cell length (mm) was measured (n450). Values are the means±s.d. (c) PC-3-shScr and PC-3-shPAK4, and DU145-shScr and DU145-shPAK4 cells were stained with Annexin-V and analyzed by ﬂow cytometry. (d) Xenograft tumors from PC-3-shScr and PC-3-shPAK4 cells were subjected to a TUNEL assay and stained for Bcl-2. Scale bar, 20 mm. (e) PC-3-shScr and PC-3-shPAK4 cells were incubated with docetaxel (50 nM), paclitaxel (100 nM), and doxorubicin (1 mM) for 3 days. Apoptotic cells were measured as described above.

suggests that anti-androgen therapy induces NE differentiation Importantly, we found that this pathway is actively involved in both in vitro and in vivo,29 and NE differentiation contributes to prostate tumorigenesis and contributes to therapy resistance. androgen-independence. Androgen depletion elevates cAMP These results provide a strong rationale for targeting PAK4 in levels,30 which leads to NE differentiation through PKA.19 The advanced prostate cancer. present study suggested that PAK4 is a PKA effector in this pathway. In addition to enhanced CREB-dependent transcription, PAK4 may contribute to the generation of neurite-like MATERIALS AND METHODS processes during NE differentiation through remodeling of Materials the actin cytoskeleton, which may involve the PAK4/LIMK/ Forskolin, dibutyryl-cAMP, H89, GF109203X, DHT, Hoechst 33,258, pacli- cofilin pathway.31 Additionally, because NE cells secrete many taxel, docetaxel and doxorubicin were purchased from Sigma-Aldrich (St cytokines and growth factors, PAK4 may indirectly regulate Louis, MO, USA). The following antibodies were purchased from Cell multiple signaling and transcription pathways through these Signaling Technology (Boston, MA, USA): phospho-Ser474 and total PAK4; phospho-Thr197 and total PKA-Ca; phospho-PKA substrate (RRXS*/T*); effects. phospho-Ser133 and total CREB; and Bcl-2. The anti-neuron-specific Overcoming chemotherapy resistance is a major goal for enolase antibody was purchased from Dako (Carpinteria, CA, USA). The current research. Our data showed that the inhibition or down- cyclin A1, Myc and GFP antibodies were purchased from Santa Cruz regulation of PAK4 potently enhances the sensitivity of prostate Biotechnology (Santa Cruz, CA, USA). Myelin basic protein (MBP) and cancer cells to anti-cancer drugs. Bcl-2 is involved in the [g-32P]ATP were purchased from Millipore (Billerica, MA) and PerkinElmer progression to androgen-independence and therapy resistance (Waltham, MA, USA), respectively. The Dual-Luciferase Reporter assay in prostate cancer,32,33 and Bcl-2 was found to be downstream of system and the fluorescein isothiocyanate Annexin-V Apoptosis Detection PAK4 in our experiments. Therefore, PAK4 may be both a valuable kit were purchased from Promega (Madison, WI, USA) and BD Biosciences marker in predicting poor prognosis, and a therapeutic target. As a (San Jose, CA, USA), respectively. Plasmids (pRS) for human scrambled (50-GCACTACCAGAGCTAACTCAGATAGTACT-30) or PAK4-specific shRNA kinase, PAK4 is a potential target for small-molecule inhibitors, (50-TGACTACTGCACCTGGACAGCCTCCTCTT-30) were purchased from Ori- which may be highly valuable for the treatment of advanced Gene Technologies (Rockville, MD, USA). Human PAK4 siRNA (50- prostate cancer. CUUCAUCAAGAUUGGCGAGUU-30) was supplied from Bioneer (Daejeon, In summary, we report the identification a novel signaling Korea). PC-3, DU145 and LNCap-FGC cells were obtained from Korean Cell pathway involving PKA, PAK4 and CREB in prostate cancer. Line Bank (Seoul, Korea).

Oncogene (2013) 2475 – 2482 & 2013 Macmillan Publishers Limited Role for PAK4 in prostate cancer progression M-H Park et al 2481 DNA constructs at room temperature (RT), and washed three times with TBS-T. Membranes Human PAK4 complementary DNA (cDNA) (MGC-9413) was obtained from were then incubated with a horseradish peroxidase-conjugated secondary the American Tissue Type Culture Collection (Rockville, MD, USA), and the antibody for 1 h at RT and were washed with TBS-T three times. The signal catalytic alpha (C, BC003238) and regulatory beta I (R, BC011424) subunits was detected using an enhanced chemiluminescence reagent. of PKA cDNA were purchased from 21C Frontier Human Gene Bank (Daejeon, Korea). PAK4 cDNAs for the full-length form (1-1776 bp) as well Immunohistochemistry as the regulatory (1-900 bp) and catalytic (901-1776 bp) domains, were Tumor xenografts were fixed with 10% neutral buffered formalin, amplified by PCR using each specific primer set, and were cloned into dehydrated, and embedded in paraffin. Sections (4 mm thickness) were pCS4-3Myc or pGEX-4T-1. The GST or Myc-tagged PAK4K350M mutants cut from the formalin-fixed, paraffin-embedded tissue blocks. After (S104A, S474A and S104A/S474A) were generated by site-specific deparaffinization, sections were subjected to an antigen retrieval mutagenesis from the wild-type of GST- or Myc-PAK4 using the procedure in 0.01 M sodium citrate buffer (pH 6.0) for 10 min, using a QuickChange Site Directed Mutagenesis Kit (Stratagene, LaJolla, CA, pressure cooker (Decloaking chamber; Biocare Medical, Concord, CA, USA). USA). The cDNA for His-tagged PKA-Ca was amplified using PCR and The sections were then incubated in methanol containing 0.3% hydrogen was cloned into the pET-24a vector. cDNAs for PKA-Ca and RI were peroxide at RT for 20 min to block endogenous peroxidase activity. After amplified by PCR and were cloned into the pEGFP-C2 vector (Clontech, blocking endogenous biotin, the sections were incubated with Protein Mountain View, CA, USA). Block Serum-Free (DAKO, Carpentaria, CA, USA) at RT, for 10 min, to block non-specific staining, and then incubated overnight at 4 1C with the anti- Generation of stable PAK4 knockdown cell lines PAK4, NSE, CREB or Bcl-2 antibody. After washing, the sections were PC-3, DU145 and LNCap-FGC cells were routinely cultured in RPMI 1640 incubated with a biotin-conjugated secondary antibody at RT for 30 min supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA, USA) and finally with peroxidase-conjugated streptavidin at RT for 30 min. Peroxidase activity was detected using the enzyme substrate diamino- and an antibiotic/antimycotic solution at 37 1C in a 5% CO2 humidified incubator. To produce stable cell lines, PC-3 and DU145 cells were plated in benzidine-HCl (DAB). For the negative controls, the sections were treated a 60-mm-diameter culture dish and were transfected with the pRS with TBS, instead of a specific primary antibody. For histologic evaluation, plasmids (5 mg) expressing the scrambled or PAK4-specific shRNA using sections were stained by H&E. Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). Cells were selected with puromycin-containing media for 2 weeks to generate single-clonal Immunocytochemistry populations. PAK4 knockdown was detected by immunoblotting with an Cells were fixed for 15 min with 3.7% paraformaldehyde and were anti-PAK4 antibody. permeabilized for 5 min with 0.2% Triton X-100. Fixed cells were blocked for 30 min at 25 1C with 2% BSA in phosphate-buffered saline. For the In vitro kinase assay staining, the cells were incubated with an anti-Myc antibody for 1 h at 25 1C, followed by incubation with an Alexa Fluor 488 conjugated Immunoprecipitated PAK4 and purified recombinant GST-PAK4 (2 mg) was secondary antibody for 1 h. To visualize F-actin, cells were stained with incubated with MBP (2 mg) and PKA-C (50 ng) (Calbiochem, LaJolla, CA, Alexa Fluor 594 conjugated phalloidin for 30 min at 25 1C. Cells expressing USA), respectively, in kinase assay buffer (20 mM Tris–HCl, pH 8.0; 10 mM Mg 32 the PAK4 constructs were observed and cell length was measured using Cl , and 1 mM dithiothreitol) containing 100 mM ATP and 5 mCi [g- P]ATP for 2 MetaMorph software version 7.1.7 (Molecular Devices, Sunnyvale, CA, USA). 30 min at 30 1C. Phosphorylation of MBP and PAK4 proteins was detected by autoradiography. TUNEL assay Tumorigenicity assay To detect apoptotic cells in tumor xenografts, TUNEL staining was performed using the ApopTag Fluorescein direct In Situ Apoptosis Five-week-old female BALB/c nude mice were purchased from Orient Bio Detection Kit (Chemicon, Billerica, MA, USA). Briefly, sections were post- (Seoul, Korea). Mice were housed in a laminar flow cabinet with a 12 h fixed in a precooled solution of ethanol: acetic acid (2:1) for 5 min and were light/dark cycle and were maintained on standard laboratory chow washed with equilibration buffer for 10 s. The slides were incubated with ad libitum. All experimental animals used in this study were maintained the working TdT enzyme in a humidified chamber at 37 1C for 1 h, and then under a protocol approved by the Institutional Animal Care and Use washed with stop/wash buffer. Finally, the slides were counter-stained with Committee at Chonbuk National University. Stable prostate cancer cells Hoechst 33258 and were mounted with Vectashield medium. (1 Â 106) expressing scrambled shRNA and PAK4 shRNA were mixed with same amount of Matrigel (BD Biosciences) and were injected subcuta- neously into the right side of the anterior flank of each mouse. The tumor Apoptosis assay size of each mouse was regularly measured with a caliper regularly and was Cells were trypsinized and stained with fluorescein isothiocyanate- calculated using the following equation; V (in mm3) ¼ width Â length Â Annexin-V and propidium iodide. Apoptotic cells were analyzed with height. Monitoring was continued until the termination of this study. On FACScan (BD Biosciences). The FL1-H value (%) was used to represent day 60, all of the mice were euthanized and tumor xenografts were excised Annexin-V-positive cells. and weighed. These tumor xenografts were then used for hematoxylin and eosin (H&E), immunohistochemistry and western blot analysis. Statistical analysis Paired t-tests were performed using SPSS version 12.0 for Windows (SPSS cAMP response element (CRE) reporter assay Inc., Chicago, IL, USA) and statistical significance was set at Po0.05 or 0.01. 293T cells were co-transfected with various Myc-PAK4 constructs or GFP- PKA-C plus the CRE-luciferase reporter. The Renilla-luciferase vector was used as an internal control. At 24 h post-transfection, luciferase activity was CONFLICT OF INTEREST measured using a Dual-Luciferase Reporter assay system. The data were The authors declare no conflict of interest. recorded as relative values of CRE activity to Renilla-luciferase.

Immunoprecipitation and immunoblotting ACKNOWLEDGEMENTS CRE-luciferase reporter gene containing 6x CRE binding sites was gift from Dr Jung EB Cells were lysed with cold lysis buffer (50 mM HEPES, pH 7.5; 150 mM NaCl, at Chungbuk National University (Cheongju, Korea). This work was supported by a 10% glycerol, 1% Triton X-100, 500 mM EDTA, 200 mM sodium-pyruvate, and grant of the National Research Foundation of Korea (2012-0005747). MAS was 50 mM b-glycerol-phosphate) and lysates were immunprecipitaed with the primary antibody at 4 1C for 18 h. Immunoprecipitates were fractionated by supported by USPHS grant RO1 CA14397. 8–10% SDS–PAGE, and were transferred to a polyvinylidene diﬂuoride membrane in a Tris-glycine-methanol buffer (25 mM Tris base, 200 mM glycine and 20% methanol). Membranes were blocked with 3% bovine REFERENCES serum albumin in Tris-buffered saline (TBS-T; 50 mM Tris, 150 mM NaCl, and 1 Jemal A, Siegel R, Xu J, Ward E. Cancer statistics 2010CA Cancer J Clin 2010; 60: 0.1% Tween-20) for 30 min, incubated with the primary antibodies for 1 h 277–300.

& 2013 Macmillan Publishers Limited Oncogene (2013) 2475 – 2482 Role for PAK4 in prostate cancer progression M-H Park et al 2482 2 Debes JD, Tindall DJ. Mechanisms of androgen-refractory prostate cancer. New 19 Bang YJ, Pirnia F, Fang WG, Kang WK, Sartor O, Whitesell L et al. Terminal neu- Engl J Med 2004; 351: 1488–1490. roendocrine differentiation of human prostate carcinoma cells in response to 3 Donkena KV, Yuan H, Young CY. Recent advances in understanding hormonal increased intracellular cyclic AMP. Proc Natl Acad Sci USA 1994; 91: 5330–5334. therapy resistant prostate cancer. Curr Cancer Drug Targets 2010; 10: 402–410. 20 Cox ME, Deeble PD, Bissonette EA, Parsons SJ. Activated 3’,5’-cyclic AMP-depen- 4 Jaffer ZM, Chernoff J. p21-activated kinases: three more join the Pak. Int J Biochem dent protein kinase is sufﬁcient to induce neuroendocrine-like differentiation of Cell Biol 2002; 34: 713–717. the LNCaP prostate tumor cell line. J Biol Chem 2000; 275: 13812–13818. 5 Bokoch GM. Biology of the p21-activated kinases. Annu Rev Biochem 2003; 72: 21 Wilson BE, Mochon E, Boxer LM. Induction of bcl-2 expression by phosphorylated 743–781. CREB proteins during B-cell activation and rescue from apoptosis. Mol Cell Biol 6 Morreale A, Venkatesan M, Mott HR, Owen D, Nietlispach D, Lowe PN et al. 1996; 16: 5546–5556. Structure of Cdc42 bound to the GTPase binding domain of PAK. Nat Struct Biol 22 Wells CM, Abo A, Ridley AJ. PAK4 is activated via PI3K in HGF-stimulated epithelial 2000; 7: 384–388. cells. J Cell Sci 2002; 115: 3947–3956. 7 Lei M, Lu W, Meng W, Parrini MC, Eck MJ, Mayer BJ et al. Structure of PAK1 in an 23 Paliouras GN, Naujokas MA, Park M. Pak4 a novel Gab1 binding partner, mod- autoinhibited conformation reveals a multistage activation switch. Cell 2000; 102: ulates cell migration and invasion by the Met receptor. Mol Cell Biol 2009; 29: 387–397. 3018–3032. 8 Qu J, Cammarano MS, Shi Q, Ha KC, de Lanerolle P, Minden A. Activated PAK4 24 Higuchi M, Onishi K, Kikuchi C, Gotoh Y. Scaffolding function of PAK in the PDK1- regulates cell adhesion and anchorage-independent growth. Mol Cell Biol 2001; Akt pathway. Nat Cell Biol 2008; 10: 1356–1364. 21: 3523–3533. 25 Baldassa S, Calogero AM, Colombo G, Zippel R, Gnesutta N. N-terminal interaction 9 Liu Y, Xiao H, Tian Y, Nekrasova T, Hao X, Lee HJ et al. The pak4 protein kinase domain implicates PAK4 in translational regulation and reveals novel cellular plays a key role in cell survival and tumorigenesis in athymic mice. Mol Cancer Res localization signals. J Cell Physiol 2010; 224: 722–733. 2008; 6: 1215–1224. 26 Shaywitz AJ, Greenberg ME. CREB: a stimulus-induced transcription factor 10 Eswaran J, Soundararajan M, Knapp S. Targeting group II PAKs in cancer and activated by a diverse array of extracellular signals. Annu Rev Biochem 1995; 68: metastasis. Cancer Metastasis Rev 2009; 28: 209–217. 821–861. 11 Wells CM, Jones GE. The emerging importance of group II PAKs. Biochemical J 27 Xiao X, Li BX, Mitton B, Ikeda A, Sakamoto KM. Targeting CREB for cancer therapy: 2010; 425: 465–473. friend or foe. Curr Cancer Drug Targets 2010; 10: 384–391. 12 Callow MG, Clairvoyant F, Zhu S, Schryver B, Whyte DB, Bischoff JR et al. 28 Murray BW, Guo C, Piraino J, Westwick JK, Zhang C, Lamerdin J et al. Requirement for PAK4 in the anchorage-independent growth of human cancer Small-molecule p21-activated kinase inhibitor PF-3758309 is a potent inhibitor cell lines. J Biol Chem 2002; 277: 550–558. of oncogenic signaling and tumor growth. Proc Natl Acad Sci USA 2010; 107: 13 Siu MK, Chan HY, Kong DS, Wong ES, Wong OG, Ngan HY et al. p21-activated 9446–9451. kinase 4 regulates ovarian cancer cell proliferation, migration, and invasion and 29 Yuan TC, Veeramani S, Lin MF. Neuroendocrine-like prostate cancer cells: neu- contributes to poor prognosis in patients. Proc Natl Acad Sci USA 2010; 107: roendocrine transdifferentiation of prostate adenocarcinoma cells. Endoc Relat 18622–18627. Cancer 2007; 14: 531–547. 14 Wells CM, Whale AD, Parsons M, Masters JR, Jones GE. PAK4: a pluripotent kinase 30 Burchardt T, Burchardt M, Chen MW, Cao Y, de la Taille A, Shabsigh A et al. that regulates prostate cancer cell adhesion. J Cell Sci 2010; 123: 1663–1673. Transdifferentiation of prostate cancer cells to a neuroendocrine cell phenotype 15 Eswaran J, Lee WH, Debreczeni JE, Filippakopoulos P, Turnbull A, Fedorov O et al. in vitro and in vivo. J Urol 1999; 162: 1800–1805. Crystal Structures of the p21-activated kinases PAK4, PAK5, and PAK6 reveal 31 Ahmed T, Shea K, Masters JR, Jones GE, Wells CM. A PAK4-LIMK1 pathway catalytic domain plasticity of active group II PAKs. Structure 2007; 15: 201–213. drives prostate cancer cell migration downstream of HGF. Cell Signal 2008; 20: 16 Lee KH, Lee S, Lee WY, Yang HW, Heo WD. Visualizing dynamic interaction 1320–1328. between calmodulin and calmodulin-related kinases via a monitoring method in 32 McDonnell TJ, Troncoso P, Brisbay SM, Logothetis C, Chung LW, Hsieh JT et al. live mammalian cells. Proc Natl Acad Sci USA 2010; 107: 3412–3417. Expression of the protooncogene bcl-2 in the prostate and its association 17 Bagchi G, Wu J, French J, Kim J, Moniri NH, Daaka Y. Androgens transduce the G with emergence of androgen-independent prostate cancer. Cancer Res 1992; 52: alphas-mediated activation of protein kinase A in prostate cells. Cancer Res 2008; 6940–6944. 68: 3225–3231. 33 Raffo AJ, Perlman H, Chen MW, Day ML, Streitman JS, Buttyan R. Overexpression 18 Gonzalez GA, Montminy MR. Cyclic AMP stimulates somatostatin gene tran- of bcl-2 protects prostate cancer cells from apoptosis in vitro and confers scription by phosphorylation of CREB at serine 133. Cell 1989; 59: 675–680. resistance to androgen depletion in vivo. Cancer Res 1995; 55: 4438–4445.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2013) 2475 – 2482 & 2013 Macmillan Publishers Limited