Hsp27 Promotes Insulin-Like Growth Factor-I Survival Signaling in Prostate Cancer via p90Rsk-Dependent Phosphorylation and Inactivation of BAD

Prostate cancer is the second leading cause of cancer deaths in North American men, killing more than 40,000 annually. A major challenge for the treatment of men with metastatic or recurrent disease is the development of treatment resistance after androgen deprivation therapy even when combined with docetaxel chemotherapy. Although the majority of patients initially respond to such treatments, most go on to develop lethal castrate-resistant prostate cancer (CRPC). Under conditions of castrate levels of serum testosterone, other growth factors assume a more dominant role in promoting cancer cell survival (1). CRPC is a complex process involving clonal selection (2) and androgen receptor (AR) activation by low levels of androgens, receptor mutations, or increased coactivator levels (3–5). Additional mechanisms include the activation of alternative growth factor pathways such as insulin-like growth factor (IGF; refs. 6, 7) and upregulation of antiapoptotic genes including Hsp27 (8).

The IGF axis is an important regulator of growth, survival, and metastatic potential in a variety of malignancies and is strongly implicated in prostatic carcinogenesis and CRPC (9, 10). IGF-I mediates growth and survival responses by binding to the tyrosine kinase IGF-I receptor (IGF-IR). Once activated, IGF-IR phosphorylates the insulin receptor substrate, which leads to the activation of many signaling pathways, including Ras/Raf/mitogen-activated protein kinase (MAPK) or PI3K/Akt. Activation of PI3K/Akt by IGF-I leads to Bad phosphorylation on Ser¹³⁶ and the activation of forkhead-regulated transcription factor (11). Activation of Erk phosphorylates p90 ribosomal S6 kinase (p90Rsk) which, in turn, leads to the phosphorylation of Bad on Ser¹¹² (12). Hence, the activation of PI3K and Erk pathways converge to inhibit apoptosis by phosphorylating (on Ser¹³⁶ and Ser¹¹², respectively) and inactivating the proapoptotic protein Bad.

Hsp27 is a stress-induced molecular chaperone that is highly and uniformly expressed in CRPC (13, 14) and other cancers (15–17), and is often associated with metastasis and poor prognosis (18, 19). Hsp27 expression is induced by hormone therapy or chemotherapy and inhibits treatment-induced apoptosis through multiple mechanisms (8, 18, 20–22). Hsp27 has been shown to interact and inhibit components of both stress-induced and receptor-induced apoptotic pathways. Hsp27 prevents the activation of caspases by sequestering cytochrome c in the cytoplasm (23). Cytochrome c interacts with Apaf-1 and caspase-9 to form the “apoptosome”, which activates caspase-3 and a cascade of downstream caspases, the so-called “effectors” of cell death (24). Hsp27 interacts with and inhibits caspase-3 activation, an effect related to the ability of Hsp27 to stabilize actin microfilaments (25). Hsp27 binds to filamentous actin to prevent the disruption of the cytoskeleton resulting from either heat shock and other stresses (26), and also inhibits the mitochondrial release of Smac to confer resistance to dexamethasone in multiple myeloma cells (27). Hsp27 is also involved in the regulation of AKT, inhibiting Bax activation to enhance cell survival (28). Hsp27 could inhibit apoptosis induced by etoposide or tumor necrosis factor-α by increasing IκBα ubiquitination and degradation, which increases the activity of the survival transcription factor nuclear factor-κB (29).

Although both increased IGF-I/IGF-IR signaling and Hsp27 expression are associated with CRPC progression (6, 7, 30, 31), the precise mechanisms by which the prosurvival effects of IGF-I/Hsp27 drive CRPC progression are incompletely defined. Here, we define the links between Hsp27 phosphorylation and IGF-I signaling in CRPC, identifying p90Rsk as a downstream effector of IGF-I–activated Erk that directly phosphorylates Hsp27 as a key step in the inhibition of Bad/14-3-3 complex formation and apoptosis.

Hsp27 wild-type was subcloned into lentivirus vector pHR-cytomegalovirus as previously described (8). Hsp27 triple mutant alanine (Ala) was generated by direct mutagenesis replacing Ser¹⁵, Ser⁷⁸, and Ser⁸², respectively, with alanine using QuikChange mutagenesis kit according to the instructions of the manufacturer (Strategene). p90Rsk wild-type and dominant negative (DN) plasmids were generously provided by Dr. J. Blenis (Harvard University, Boston, MA).

IGF-I was purchased from Research Diagnostics, Inc. SB 203580, Ro 31-8220, PD 98056, and cycloheximide were from Calbiochem. Total Hsp27/pHsp27 was from Assay Designs; pAkt/Akt, Bad, total Erk, and poly(ADP ribose)polymerase were from Cell Signaling Technology; p90Rsk/p-p90Rsk, pErk, p38 kinase, AR, IGF-IR, 14-3-3 antibodies, and p90Rsk short interfering RNA (siRNA) were from Santa Cruz Biotechnology, Inc. Purified recombinant nonphosphorylated Hsp27 (expressed in bacteria) and active p90Rsk (expressed in Sf21 cells) were purchased from Assay Designs and Upstate Biotechnology, respectively. Secondary antibodies FITC and Rhodamine Red-x were purchased from Jackson ImmunoResearch Laboratories, Inc., and Flag-Agarose was purchased from Sigma Aldrich.

All animal experimentation was conducted in accordance with accepted standards of the University of British Columbia Committee on Animal Care. LNCaP and PC-3 xenograft tumors were grown in athymic nude mice at four sites as modified from a previously reported method (5). Prostate specific antigen levels were measured weekly by tail vein sera samples using an immunoassay kit (ClinPro). Mice were castrated at 6 wk postinoculation with LNCaP cells. Tumors were harvested from the same mouse precastration (intact), and at 35 d postcastration (CRPC). Tumors were excised, dissected, and immediately frozen in liquid nitrogen.

A tissue microarray was constructed using a Beecher microarray from paraffin-embedded specimens with matching H&E slides. Each case was represented with three cancer cores in the tissue microarray analysis. Sections were deparaffinized and rehydrated with xylene and ethanol, and then transferred to 0.02% triton for permeabilization. Slides in citrate buffer (pH 6) were heated in the steamer for 30 min, washed for 5 min in PBS, incubated in 3% bovine serum albumin transferred to 3% H₂O₂, and incubated overnight with anti-pHsp27 antibodies (Cell Signaling) as described (8).

PC-3 cells were plated in DMEM with 5% fetal bovine serum. The next day, serum-free medium was used with or without 50 ng/mL of IGF-I and cell proliferation was done (32).

PC-3 cells were first serum-starved overnight in DMEM. IGF-I was then added at different concentrations or at different times. Inhibitors were preincubated for 1 h prior to the addition of IGF-I. After treatment, cells were lysed in radioimmunoprecipitation assay buffer and submitted to Western blot as previously described (32).

One microgram of recombinant Hsp27 was incubated at 30°C for 1 h with different concentrations of the purified active form of p90Rsk in reaction buffer [50 mmol/L Hepes (pH 7.4), 10 mmol/L MgCl₂, 1 mmol/L EGTA, 1 mmol/L DTT, 10 mmol/L β-mercaptoethanol, 0.5 mmol/L vanadate, 50 μmol/L ATP, and 5 uCi of ³²P-γ-ATP]. The p90Rsk inhibitor (Ro-31-8220) was added to show the specificity of the reaction.

Total proteins (500 μg) were precleared with protein-G sepharose (Invitrogen-Life Technologies, Inc.) for 1 h at 4°C and immunoprecipitated with 2 μg of anti-Bad overnight at 4°C. The immune complexes were recovered with protein-G sepharose for 2 h and then washed with radioimmunoprecipitation assay buffer at least three times, centrifuged, and submitted to SDS-PAGE followed by Western blotting. For p90Rsk pulldown, cells were lysed in radioimmunoprecipitation assay buffer 48 h after transfection and 750 μg of protein were incubated with Flag-Agarose for 3 h; the immune complexes were washed and submitted to Western blotting.

PC-3 cells were treated once daily for 2 d with 10 nmol/L of siRNA or Scr control. After the second transfection, cells were incubated ±IGF-I for 48 h. Caspase-3 assay was performed according to the recommendations of the manufacturer (BIOMOL Research Laboratories, Inc.). Briefly, 50 μg of total proteins were assessed to cleave the Ac-DEVD-AMC substrate. The cleaved products yielded a fluorogenic reaction and were read in a microplate reader using a filter with a 360 nm excitation/460 emission wavelength.

PC-3 cells were grown on coverslips and fixed in paraformaldehyde for 10 min at room temperature and immunofluorescence was performed as described previously (32) using Hsp27 (1:500) and p90Rsk (1:100) antibodies.

PC-3 cell lines were plated in 10 cm dishes and 24 h later, cells were serum-starved and treated with or without 50 ng/mL of IGF-I or 10 μg/mL of cycloheximide for 24 h. Sub-G₀ was analyzed as described previously in ref. (32).

All data were analyzed by Student's t test. Levels of statistical significance were set at P < 0.05 (two-sided), and all statistical calculations were done using StatView 4.5 (Abacus Concepts, Inc.).

To characterize the levels of both total and phosphorylated Hsp27 (p-Hsp27) levels with CRPC, we analyzed staining patterns in human prostate cancer tissues before and after androgen ablation therapy. Figure 1A illustrates that total and p-Hsp27 levels are low in untreated tumors but the staining intensity increases after androgen ablation, becoming uniformly strong in CRPC and statistically higher compared with untreated patients (Fig. 1B). The increased levels of p-Hsp27 correlated with total Hsp27. We next characterized changes in total and phosphorylated Hsp27 levels in the human LNCaP xenograft model, which mimics the progression to CRPC after castration. Western blot analysis of treatment-naïve and CRPC LNCaP tumors showed that p-Hsp27 levels increased after castration with no significant difference in total Hsp27 levels (Fig. 1C). This increase in p-Hsp27 levels correlated with increased levels of phosphorylated (but not total) Erk/p90Rsk and Akt protein levels (Fig. 1C). Increased p-Hsp27 levels also correlated with increased prostate serum antigen levels, suggesting a correlation between Hsp27 phosphorylation and AR-regulated prostate serum antigen expression as reported previously (32). Next, we evaluated p-Hsp27 levels in androgen-responsive LNCaP cells, castrate-resistant C4-2 cells, and AR-negative PC-3 cells. Western blot analysis indicates that whereas total Hsp27 levels are slightly higher in PC-3 xenografts, p-Hsp27 levels are highly upregulated in PC-3 tumors compared with treatment-naïve LNCaP tumors (Fig. 1D). Hsp27 expression and phosphorylation levels were also higher in C4-2 compared with LNCaP cells (data not shown). Collectively, these data indicate that increased p-Hsp27 levels correlate with CRPC.

We selected PC-3 cells to study the effect of IGF-I on cell proliferation and survival because PC-3 cells are IGF-IR positive, AR negative, and express high basal levels of p-Hsp27 (Fig. 2A). IGF-I increased proliferative (Fig. 2B, left) and decreased apoptotic (Fig. 2B, right) rates in PC-3 cells grown in serum-free medium. Next, we investigated the effect of IGF-I on PC-3 cell survival using the apoptotic inducer cycloheximide. As expected, cycloheximide treatment increased cleaved poly(ADP ribose)polymerase (Fig. 2C, left) and sub-G₀ population (Fig. 2C, right) as indicators of apoptosis, which were both reduced in the presence of IGF-I. Together, these results indicate that IGF-I is cytoprotective in PC-3 cells.

To identify the relationships between Hsp27 in IGF-I signaling and cell survival in prostate cancer, we next determined if IGF-I phosphoactivates Hsp27. A dose-dependent and time course analysis of p-Hsp27 after IGF-I treatment was studied in parallel with the activation of IGF-IR signaling pathway. PC-3 cells were serum-starved overnight and treated with increasing concentrations of IGF-I. IGF-I potently induced dose- and time-dependent increases in Hsp27, Akt, and Erk phosphorylation (Fig. 2D, left). Interestingly, p-Hsp27 increased within 5 minutes after IGF-I exposure (Fig. 2D, middle), and as expected, IGF-I stimulated the time-dependent phosphorylation of p38 kinase, Akt, Erk, and p90Rsk. Moreover, IGF-I induced Bad/14-3-3 interaction in a time-dependent manner (Fig. 2D, right) as a consequence of Bad phosphorylation by both Akt and Erk pathways. These results confirm that IGF-I phosphoactivates Hsp27 and enhances Akt and Erk signaling in PC-3 prostate cancer cells.

To identify upstream effectors of IGF-I–induced Hsp27 phosphorylation, we pretreated PC-3 cells prior to IGF-I stimulation with selected specific inhibitors, SB 203580, LY 294002, and PD 98059 targeting p38 kinase, Akt, and Erk, respectively. The phosphorylation status of Hsp27, p38 kinase, Akt, and Erk were analyzed by Western blotting. Although neither SB203580 nor LY294002 altered IGF-I–induced Hsp27 phosphorylation (Fig. 3A and B), the Erk inhibitor PD 98059 abolished IGF-I–induced Hsp27 phosphorylation (Fig. 3C). We next tested whether IGF-I–induced Hsp27 phosphorylation occurs directly via Erk kinase or via its downstream effector, p90Rsk. PC-3 cells treated with the Ro-31-8220 inhibitor completely abolished IGF-I–induced Hsp27 phosphorylation (Fig. 3D), suggesting that IGF-I–mediated phosphorylation of Hsp27 occurs via the MAPK pathway downstream of Erk kinase.

To determine whether p90Rsk phosphorylates Hsp27, we performed an in vitro kinase assay using purified unphosphorylated Hsp27 and constitutively active p90Rsk. Figure 4A shows that in the absence of p90Rsk, Hsp27 is unphosphorylated and that only 50 ng of p90Rsk was sufficient to phosphorylate 1 μg of Hsp27 (1:50; Fig. 4A, left). Moreover, the p90Rsk inhibitor Ro-31-8220 significantly suppressed the ability of p90Rsk to phosphorylate Hsp27 (Fig. 5A, right), suggesting that p90Rsk specifically and directly phosphorylates Hsp27 in vitro. To corroborate this finding, we examined the ability of p90Rsk to phosphorylate Hsp27 in vivo. Overexpression of wild-type p90Rsk increased Hsp27 phosphorylation, which was amplified in the presence of IGF-I stimulation compared with empty vector. However, p90Rsk dominant negative/kinase dead overexpression decreases endogenous Hsp27 and p90Rsk phosphorylation in the absence or presence of IGF-I stimulation (Fig. 4B, left). To further support these results, p90Rsk siRNA decreased p-Hsp27 levels that directly correlated with p90Rsk knockdown (Fig. 4B, right). Interestingly, p90Rsk interacts with Hsp27 (Fig. 4C, left), and this interaction is dependent on p90Rsk activation because the DN of p90Rsk abolished p90Rsk/Hsp27 interaction (Fig. 4C, right). Hsp27 and p90Rsk colocalize in the cytoplasm in PC-3 cells (Fig. 4D, left). Moreover, immunohistochemical analysis of human prostate cancer specimens indicates that Hsp27 and p90Rsk are present in the same subset of the cells after androgen ablation (Fig. 4D, right). Collectively, these results indicate that p90Rsk directly interacts with and phosphorylates Hsp27 in vitro and in vivo, and that this interaction is p90Rsk-dependent.

Akt and Erk activation leads to phosphorylation of Bad at Ser¹¹² and Ser¹³⁶, which increases 14-3-3 binding and sequesters Bad in the cytoplasm to suppress its proapoptotic activity (33). Because Hsp27 is a cytoprotective chaperone and is phosphorylated by p90Rsk, and Hsp27 knockdown increases apoptosis rates, we hypothesized that Hsp27 levels may modulate cell survival via the modulation of Bad phosphorylation and formation of the Bad/14-3-3 complex. We first characterized the effects of Hsp27 on the levels of IGF-IR downstream targets after IGF-I stimulation and found that Hsp27 knockdown abolished IGF-I–induced phosphorylation of Hsp27, Akt, Erk, and Bad on Ser¹¹² and Ser¹³⁶ without affecting their total expression levels (Fig. 5A). In contrast, Hsp27 wild-type overexpression enhances the phosphorylation of Akt, Erk, and p90Rsk compared with empty vector (Fig. 5B). An Hsp27 triple mutant (Ser¹⁵, Ser⁷⁸, and Ser⁸² were mutated with alanine) also decreased the levels of phosphorylated Akt, Erk, and p90Rsk, consistent with its dominant-negative effect and loss of Hsp27 function (Fig. 5C). Next, we investigated the effect of Hsp27 on Bad phosphorylation as a downstream effector of Akt and MAPK pathways. Overexpression of Hsp27 wild-type increases the phosphorylation of Bad and Bad/14-3-3 complex formation, whereas Hsp27 downregulation at expression (siRNA) or phosphorylation (triple mutant) levels decreased the levels of Bad phosphorylation on Ser¹¹² and Ser¹³⁶ and Bad/14-3-3 complex, even in the presence of IGF-I. We also found that silencing Hsp27 using OGX-427 (data not shown) or siRNA in PC-3 cells treated with or without IGF-I increased the sub-G₀ fraction (Fig. 6A) and caspase-3 activity (Fig. 6B). These results show that Hsp27 silencing increases apoptotic rates and this effect was not reversed by IGF-I. Collectively, these results indicate that the levels of total and p-Hsp27 are important determinants of Bad/14-3-3 complex stability and represent one mechanism regulating cancer cell survival.

Survival proteins and signaling pathways upregulated following androgen ablation that function to inhibit cell death are of special importance to treatment resistance and the progression of CRPC. Increased activity in the IGF-IR signaling pathway and altered expression of IGF-I axis components are strongly implicated in prostate carcinogenesis and CRPC progression (30, 34), and contributes to nearly 50% of cell growth and proliferation in some models (35). Castrate-resistant progression of human prostate cancer xenografts is associated with increased expression of IGF-IR and IGF-I signaling pathways (30, 36). In prostate cancer and other cancer models, IGF-I is mitogenic and antiapoptotic, and IGF-IR blockade inhibits castration-induced prostate cancer xenograft growth (37). The PI3/Akt and Erk signaling pathways are the principal cell survival pathways activated by the IGF-I axis (38), and include the phosphorylation, and subsequently, deactivation of the proapoptotic Bad protein (39). Hsp27 is phosphoactivated during cell stress to form oligomers that prevent protein aggregation and/or regulate activity/degradation of certain client proteins (40). Hsp27 phosphorylation is catalyzed by MAPKAP kinase-2, a target of the p38 MAPK (41, 42). Higher levels of Hsp27 are commonly detected in many cancers (18) in which it plays a role in thermotolerance, cytoprotection, steroid hormone response, and molecular chaperoning (43). Hsp27 is also involved in the regulation of Akt (44) and enhances nuclear factor-κB activity by increasing the degradation of IκBα (29). Hsp27 also chaperones and shuttles stat3 and AR into the nucleus, enhancing the transcription of several Stat3- and AR-regulated survival genes (8). In prostate cancer, stress-induced increases in Hsp27 after hormone therapy inhibit castration-induced cell death, accelerate CRPC progression, and render cells more resistant to chemotherapy (8, 19).

Because Hsp27 is involved in many pathways regulating the response of a cell to stress and therapeutic stimuli, including MAPK and Akt pathways, and that increased Hsp27 and IGF-I activity are both associated with CRPC progression (19, 30), we explored the interactions between IGF-I signaling and Hsp27 phosphoactivation that could promote castrate-resistant disease. Here, we report that both total and p-Hsp27 are highly expressed in and correlates with CRPC in human prostate tissues (Fig. 1A) and LNCaP xenografts postcastration (Fig. 1B). Increased p-Hsp27 levels are accompanied by Akt, Erk, and p90Rsk phosphorylation, all involved in cell survival and proliferation. We previously reported that Hsp27 phosphorylation plays a key role in AR activation and stability (32), that increased Hsp27 levels accelerate castrate-resistant LNCaP xenograft growth, and that Hsp27 knockdown using antisense (OGX-427; OncoGenex Technologies) or siRNA delays progression postcastration. Collectively, these data functionally link phosphoactivated Hsp27 to stress-induced prostate cancer cell survival and castrate-resistant progression (8, 32).

Depending on cell context and stimulus, Hsp27 phosphorylation is regulated by several protein kinases involved in cell survival and proliferation, including p38 kinase and Akt (44, 45). PKCα, PKCδ, and cyclic AMP–dependent kinase have also been reported to phosphorylate Hsp27 (46), suggesting that Hsp27 regulates the activity of many varied signaling pathways. We show here for the first time that Hsp27 is directly phosphorylated by p90Rsk kinase, a downstream effector of MAPK pathway, in an IGF-I–dependent manner. Several lines of evidence support a link between Hsp27 phosphorylation and p90Rsk in prostate cancer progression. First, both Hsp27 and p90Rsk are overexpressed in prostate cancer and both enhance AR transcription activity (32, 47). Using in vitro and in vivo assays, we show that p90Rsk directly phosphorylates Hsp27 and that dominant negative p90Rsk inhibits Hsp27 phosphorylation. Immunoprecipitation and immunofluorescence studies indicate that Hsp27 is associated and colocalized with p90Rsk. Moreover, Hsp27 and p90Rsk are present in the same subsets of the cells in prostate cancer tumors after androgen ablation. Collectively, these studies indicate that p90Rsk interacts and phosphorylated Hsp27, a key functionally relevant mechanism linked to CRPC growth and survival.

p-Hsp27 is a downstream effector of several signaling pathways linked to cell survival and CRPC including MAPK/Erk, PI3K/Akt, and p38 kinase. In the LNCaP xenograft model, increased p-Hsp27 levels correlate with increased levels of phosphorylated Erk, p90Rsk, and Akt, suggesting that growth factor signaling pathways like IGF-I and interleukin-6 are activated after castration. Enhanced IGF-I signaling, mediated by increased IGF-IR (30), IGFBP-2 (48), and IGFBP-5 (7) levels, is believed to be a functionally and clinically relevant driver of CRPC progression (49). Interactions between IGF-I signaling and Hsp27 via activation of MAPK/p90Rsk and Hsp27 phosphoactivation lead to Bad/14-3-3 complex stabilization, Bad sequestration in the cytoplasm, and inhibition of apoptosis. In summary, Hsp27 is a key interacting component of the IGF-I signaling in the development of CRPC. Hsp27 is directly phosphorylated by p90Rsk to serve as a downstream effector of IGF-I signaling, promoting cell survival pathways by chaperoning and stabilizing Bad/14-3-3 complexes. These data support the development of targeted inhibitors of Hsp27 in CRPC.

The University of British Columbia has submitted patent applications, listing M.E. Gleave as inventor, on the antisense sequence described in this article. This IP has been licensed to OncoGenex Pharmaceuticals, a Vancouver-based biotechnology company in which M.E. Gleave has founding shares.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.