Heat Shock Protein 27 Confers Resistance to Androgen Ablation and Chemotherapy in Prostate Cancer Cells Through Eif4e

Home , Hsp27

Oncogene (2010) 29, 1883–1896 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 $32.00 www.nature.com/onc ORIGINAL ARTICLE Heat shock protein 27 confers resistance to androgen ablation and chemotherapy in prostate cancer cells through eIF4E

C Andrieu1,2,6, D Taieb1,2,6, V Baylot1,2, S Ettinger3, P Soubeyran1,2, A De-Thonel4, C Nelson3, C Garrido4,ASo3, L Fazli3, F Bladou5, M Gleave3, JL Iovanna1,2 and P Rocchi1,2

1INSERM, U624 ‘Stress Cellulaire’, Marseille, France; 2Aix-Marseille Universite´, Campus de Luminy, Marseille, France; 3The Prostate Centre, Vancouver General Hospital, Vancouver, British Columbia, Canada; 4INSERM U866, Faculte´ de Me´decine, Dijon, France and 5Service d’Urologie de l’Hoˆpital Sainte-Marguerite, Marseille, France

One strategy to improve therapies in advanced prostate Oncogene (2010) 29, 1883–1896; doi:10.1038/onc.2009.479; cancer (PC) involves targeting genes that are activated by published online 18 January 2010 androgen withdrawal to delay the emergence of the androgen-independent (AI) phenotype. Heat shock protein Keywords: prostate cancer; androgen independence; 27 (Hsp27) expression becomes highly upregulated in PC heat shock protein 27; eukaryotic translational initiation cells after androgen withdrawal or chemotherapy, in which factor; ubiquitination it functions as a cytoprotective chaperone to confer broad- spectrum treatment resistance. The purpose of this study is to elucidate anti-apoptotic pathways regulated by Hsp27 that are activated during PC progression. Using Introduction two-hybrid experiment, we found that Hsp27 was having a major role in the protein translational initiation process. Prostate cancer (PC) is one of the most common Furthermore, using complementary DNA (cDNA) micro- malignancies in industrialized countries, and the second array analysis, 4E binding protein 1 was identified as leading cause of cancer-related death in the United being proportionately and highly regulated by Hsp27. States. Although patients with localized disease may be These data led us to analyze the protein synthesis treated with surgery or radiation, androgen ablation is initiation pathway, which is a prerequisite for cell growth usually the initial therapy in patients with advanced or and proliferation. Using northern and western blot metastatic disease. Although most patients initially analysis, we found that Hsp27 downregulation decreased respond well to androgen ablation, their tumors eukaryotic translation initiation factor 4E (eIF4E) ultimately become unresponsive and recur within 2 expression at the protein, but not mRNA, level. The years as castration-refractory prostate cancer (CRPC) cytoprotection afforded by Hsp27 overexpression was (Fusi et al., 2004). In the past, chemotherapy had only a attenuated by eIF4E knockdown using specific eIF4E palliative role for men with PC and failed to produce a short interfering RNA (siRNA). Co-immunoprecipitation survival advantage or any significant measurable disease and co-immunofluorescence confirmed that Hsp27 colo- response. Recently, docetaxel-based regimens have calizes and interacts directly with eIF4E. Hsp27-eIF4E shown improved survival in men with CRPC in two interaction decreases eIF4E ubiquitination and proteaso- large, phase III studies (Petrylak et al., 2004; Tannock mal degradation. By chaperoning eIF4E, Hsp27 seems to et al., 2004). However, the median overall survival was protect the protein synthesis initiation process to enhance prolonged for only B2–3 months, and thus development cell survival during cell stress induced by castration or of novel therapeutic approaches are essential, including chemotherapy. Forced overexpression of eIF4E induces antisense and drug therapies that target relevant signaling resistance to androgen-withdrawal and paclitaxel treat- pathways (Gallagher and Gapstur, 2006). ment in the prostate LNCaP cells in vitro. These findings Expression of heat shock protein 27 (Hsp27) is identify Hsp27 as a modulator of eIF4E and establish a upregulated by hormonal ablation and chemotherapy potential mechanism for the eIF4E-regulated apoptosis and is associated with CRPC (Cornford et al., 2000; after androgen ablation and chemotherapy. Targeting Rocchi et al., 2004b). We recently showed using tissue Hsp27–eIF4E interaction may serve as a therapeutic microarray analysis in 232 specimens from hormone- target in advanced PC. naive and post-hormone-treated cancers that Hsp27 expression is low or absent in untreated PC, but starts Correspondence: Dr P Rocchi, Inserm U624, Head of the Prostate increasing 4 weeks after androgen ablation, to become Cancer Research Project, Parc Scientifique et Technologique de uniformly highly expressed in CRPC (Rocchi et al., Luminy, 163 Avenue de Luminy, Marseille, Provence 13009, France. 2005). Hsp27 is an adenosine triphosphate-independent E-mail: [email protected] molecular chaperone that is highly induced during stress 6These authors contributed equally to this work. Received 9 December 2008; revised 15 November 2009; accepted 19 responses and forms oligomers to interact with a wide November 2009; published online 18 January 2010 variety of client proteins, thus preventing their aggregation. Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1884 It has a central role in cellular signaling, as it is essential We next compared the rate of AI progression in for maintaining the activity of key signaling factors, LNCaP tumors in vivo after combined treatment with including steroid hormone receptors (Miller et al., 2005). castration and chemotherapy (defined by increased The cytoprotective effects of Hsp27 result from its tumor volume and rising PSA after castration) chaperone function, direct interference with caspase in LNCaP-Hsp27 (n ¼ 10) and control-transfected activation, modulation of oxidative stress and regulation (n ¼ 10) tumors. By 1 week after castration, LNCaP- of the cytoskeleton (Liang, 2000; Parcellier et al., 2003). Hsp27 tumors showed rapid tumor growth and rise in As an important regulator of cell survival and treatment PSA, consistent with more rapid AI tumor growth stress at many different points along the apoptotic compared with mock-transfected controls. At the time pathways, Hsp27 is now recognized as an important of killing, tumor volume (Figure 1c) was 4.3-fold higher therapeutic target. Recently, Hsp27 antisense oligodeox- in LNCaP-Hsp27 group (1214±321 mm3) than ynucleotides (ASO) and short interfering RNA (siRNA) in LNCaP-Mock control group (357±39.25 mm3 that target the human translation initiation site were **Pp0.01), and serum PSA (Figure 1d) was 9.2-fold reported to potently inhibit Hsp27 expression in human higher in LNCaP-Hsp27 group (68.5±23.14) compared prostate PC-3 cells with increased caspase-3 cleavage, with LNCaP-Mock control group (7.4±1.8 ng/ml, apoptosis and 87% suppression of cell growth (Rocchi **Pp0.01). Immunostaining performed in LNCaP- et al., 2004b, 2005, 2006). A second-generation ASO Mock and LNCaP-Hsp27 cells treated with paclitaxel targeting Hsp27 (OGX-427) is currently being tested in a 10 nM in serum-free media for 24 h showed signifi- phase I/II clinical trial for prostate, bladder, ovarian, cantly higher Ki67 (Figure 1e) and lower single-strand breast and lung cancers in Canada and the United States DNA (Figure 1f) expression level in LNCaP-Hsp27 (Hotte et al., 2009; http://www.oncogenex.ca/). compared with LNCaP-Mock control cells. Accord- An improved understanding of the mechanisms of ing to previous work published by Garrido and action of Hsp27, and characterization of additional Parcellier (Garrido et al., 2003; Parcellier et al., 2006), putative cell survival proteins, will improve our under- these data imply that increased Hsp27 levels protect standing of apoptotic regulatory pathways and androgen-dependent prostate cancer cells from treat- drug resistance. The objective of this study was to assess ment stress induced by increasing tumor proliferation the effect of Hsp27 overexpression on chemotherapy and decreasing apoptosis, thereby facilitating tumor and androgen withdrawal-induced apoptosis. We progression. chose the LNCaP tumor model that closely mimics androgen-independent (AI) progression in humans Hsp27 expression affects levels of the eIF4E, and its and produces prostate-specific antigen (PSA). LNCaP binding protein, 4E-BP1 tumors are androgen dependent when injected into male To explore the molecular mechanisms of apoptotic immunodeficient mice, and develop non-androgen- resistance associated with Hsp27 overexpression in regulated PSA gene expression after castration, LNCaP cells, we found using western blot that Hsp27 as a surrogate end point of AI progression (Gleave inhibition by OGX-427 could inhibit several proteins et al., 1992). As part of our ongoing analyses for involved in AI progression, such as androgen receptor, identifying key pathways mediating AI progression, we heat shock proteins 70 and 90 (Hsp70 and Hsp90) and overexpressed Hsp27 for gain-of-function analyses and heat shock factor 1, without affecting their mRNA level. used OGX-427 for loss-of-function analyses to under- On the other hand, we performed Hsp27 two-hybrid stand how Hsp27 imposes its cytoprotective effect experiments to find Hsp27 protein partners and found on cells after androgen withdrawal and chemotherapy that Hsp27 was having a major role in the protein in PCs. translational initiation process (not shown). Further- more, we also followed an unbiased approach using expression profiling printed with 13 791 (70 mer) human Results oligos in duplicate onto amine-coated slides purchased from Qiagen Operon (Alameda, CA, USA). Log2 plots Overexpression of Hsp27 inhibits apoptosis induced by from comparative microarray analyses of LNCaP- paclitaxel and androgen withdrawal and enhances AI Hsp27 and LNCaP-Mock cells identified the eukaryotic tumor growth translational initiation factor (eIF4E) binding protein 1 To study the effect of Hsp27 overexpression on (4E-BP1) as the top one over-expressed gene (B3 fold) combined apoptosis induced by androgen withdrawal in LNCaP-Hsp27. Conversely, Hsp27 knockdown in and chemotherapy, LNCaP-Hsp27 and LNCaP-Mock LNCaP cells treated with OGX-427 identified that cells were plated in 10 cm plates, and then treated with the most downregulated gene (B70%) was 4E-BP1 indicated concentrations of paclitaxel in serum-free (Figure 2a). Validation experiments using western blot (mimics androgen withdrawal in vitro) media for 24 h. analysis confirmed that 4E-BP1 protein levels were Crystal violet (Figure 1a) and flow cytometry directly proportional to Hsp27 levels, that is, increasing (Figure 1b) analyses indicated that LNCaP-Hsp27 were in LNCaP-Hsp27 overexpressing cells and decreasing in significantly more resistant to apoptosis induced by LNCaP cells treated with OGX-427 (Figure 2b). 4E-BP1 paclitaxel and androgen withdrawal than LNCaP-Mock is an eIF4E-binding protein that has a critical role cells (*Pp0.05, **Pp0.01, ***Pp0.001). in protein synthesis control mediated by mammalian

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1885 Crystal Violet analysis Flow Cytometry analysis 120 350 *** ** *** 300 100 ** ** 250 *** 80 *** 200 60 150 * 40 100 LNCaP-Mock LNCaP-Mock 20 50 LNCaP-Hsp27 LNCaP-Hsp27

Cell viability (%age of control) 0

0 Sub G0 phase (%age of control) 0 0.5 1 5 10 0 0.5 1 5 10 Paclitaxel concentrations (nM) Paclitaxel concentrations (nM)

2000 100 LNCaP-Mock LNCaP-Mock ** ) 3 LNCaP-Hsp27 ** LNCaP-Hsp27 1500 75 Paclitaxel ** ** Paclitaxel ** 1000 ** 50 ** ** ** PSA (ng/ml) * * 500 25 ** ** Tumor Volume (mm * ** * * *** 0 0 0 5 10 15 0 5 10 15 Weeks post-castration Weeks post-castration

Proliferation index Apoptotic index 200 120 180 *** 100 160 140 80 120 ** 100 60 80 40 60 40 20 20 Ki67 Highly positive labeled cell number (%age of control) 0 ssDNA Highly positive labeled cell number (%age of control) 0 LNCaP-Mock LNCaP-Hsp27 LNCaP-Mock LNCaP-Hsp27 Figure 1 Hsp27 overexpression increases LNCaP cell survival after combined androgen withdrawal and paclitaxel-chemotherapy treatment. After 24 h of treatment with indicated concentrations of paclitaxel in serum-free media (mimics androgen withdrawal in vitro), cell viability was determined using crystal violet (a) and apoptotic rates using flow cytometry (b). For in vivo studies, 10 nude mice were injected with LNCaP-Hsp27 or LNCaP-Mock cells (106 cells). The mice were castrated at PSA level of 40–60 ng/ml. From days 7 to 14 and 21 to 28, 0.5 mg of micellar paclitaxel was administrated by intravenous injection three times per week. (c) Tumor volume were monitored after castration, and the mean LNCaP-Hsp27 tumor volume was significantly greater than the LNCaP-Mock tumor volume by 1 week after castration, as indicated by the *symbol. (d) The mean PSA value for LNCaP-Hsp27 group was significantly greater than the LNCaP-Mock group at 6 weeks after castration, indicating a more rapid time to AI. Points, mean tumor volume in each experimental group containing 10 mice; bars, s.e.. *, **, and *** differ from LNCaP-mock control (Pp0.05; Pp0.01; and Pp0.001, respectively) using the Student’s t-test. (e) Detection of proliferation index using antibody Ki67 (Thermo Scientific) that reacts specifically with nuclear protein Ki67. (f) Detection of apoptotic cells using antibody MAB3299 (Chemicon) that reacts specifically with single-stranded DNA. Proliferation and apoptotic index were determined after counting the number of total labeled cells divided by the total cell number in the slide. The experiment was repeated in triplicate. ** and *** differ from LNCaP-Mock control (Pp0.01 and Pp0.001, respectively) using the Student’s t-test. target of rapamycin (mTOR) pathway, and hence, when compared with low-grade tumors, and that nearly in cell survival and proliferation (Heesom et al., 2001; all the 4E-BP1 proteins were phosphorylated (Armengol Topisirovic et al., 2004; Petroulakis et al., 2006) through et al., 2007; Rojo et al., 2007). the eIF4E. Recently, it has been described that total These results led us to analyze the mTOR signaling 4E-BP1 expression increases in high-grade breast tumors pathway to see whether Hsp27 could regulate the

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1886

U46922 fragile histidine triad gene FHIT

AK001667 hypothetical protein FLJ10805

NM-007338 deleted in lung and esophageal cancer 1 DLEC1 U31382 quanine nucleotide binding protein 4 GNG4 L36055 Human 4E-binding protein 1 mRNA, complete cds

X07994 lactase LCT X63755 keratin cuticle ultrahigh sulphur 1 KRN1 NM-0133267 breast cell glutaminase GA D28137 bone marrow stromal cell antigen 2 BST2 NM-016139 16.7Kd protein LOC51142

M23892 arachidonate 15-lipoxygenase ALOX15 M16707 H4 histone family-member N LNCaP-Hsp27 LNCaP OGX-427

ASO Control 70nM OGX-427 70nM LNCaP-MockLNCaP-Hsp27

anti-Hsp27 IB 27 kDa

anti-4E-BP1 IB 15-20kDa

anti-eIF4E IB 25kDa

anti-GAPDH IB 40kDa

Figure 2 Hsp27 expression level regulates the eukaryotic translational signaling pathway. (a) The effect of Hsp27 overexpression versus knockdown using OGX-427 on gene expression profile in LNCaP cells. RNA was extracted from LNCaP-Hsp27 or LNCaP- Mock and LNCaP cells 2 days after OGX-427 or ASO control treatment. Gene expression profile was performed using cDNA microarray in triplicate. This figure shows that decreased Hsp27 expression using OGX-427 correlated with a significantly decreased expression of 4E-BP1, the eukaryotic translational initiation factor binding protein activated by mammalian target of rapamycin (mTOR) signaling pathway. Pp0.05 using the Student’s t-test. (b) Proteins from LNCaP-Mock versus LNCaP-Hsp27 and LNCaP cells treated with OGX-427 or ASO control were extracted from cultured cells. Hsp27, 4E-BP1, eIF4E and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) protein levels were analyzed using immunoblotting (IB).

translational initiation factor eIF4E through its binding of Hsp27 increases the protein expression level of eIF4E protein 4E-BP1. We found that the effect of Hsp27 (Figure 2b) without any change on its mRNA expres- upon the level of eIF4E is independent of 4E-BP1 sion level (data not shown). eIF4E has been described to (Figure 6b), but Hsp27 could exert an effect directly on regulate cell survival of many human cancers and was eIF4E and we recently found that forced overexpression recently recognized as a therapeutic target in several

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1887 cancers (Mamane et al., 2004; Averous and Proud, 2006; by the ladder of high-molecular-weight species, which Petroulakis et al., 2006; Graff et al., 2008), including is a characteristic of polyubiquitinated protein. advanced PC (Graff et al., 2009). In contrast, knockdown of Hsp27 in LNCaP cells treated with OGX-427 increases eIF4E ubiquitination and decreases total eIF4E protein levels, suggesting that Cytoprotection induced by Hsp27 seems to involve eIF4E after Hsp27 knockdown, eIF4E becomes ubiquitinated In most systems, cap-dependent translation depends on and degraded by the ubiquitin–proteasome pathway eIF4E levels, and eIF4E can exert an effect as an (Figure 5b). We next examined the possible relation- oncogene when overexpressed (Mamane et al., 2004), ship between eIF4E ubiquitination and androgen facilitating malignant transformation and lymphoma- withdrawal and chemo-resistance. Target proteins are genesis in rodent cells (Ruggero et al., 2004; Wendel ubiquitinated on lysine (Lys) residues, and eIF4E et al., 2004). Knockdown of eIF4E using siRNA inhibits contains 16 Lys residues. As Murata and Shimotohno growth of several cell lines (Oridate et al., 2005). The (2006) showed that restoration of a single Lys, Lys-159, cytoprotection induced by Hsp27 overexpression in restored the ubiquitination of eIF4E to the same LNCaP cells seems to involve eIF4E activity, as eIF4E extent as wild type (wt), the effect of overexpression of knockdown using siRNA (Figure 3a) reverses the FLAG-tagged eIF4E ubiquitination mutant K159R cytoprotection to androgen withdrawal (serum-free protein in LNCaP cells (Figure 5c) was assessed media) and paclitaxel treatment normally conferred by after treatment stress. As shown in Figure 5, FLAG- Hsp27 overexpression (Figure 3b). Conversely, over- tagged eIF4E ubiquitination mutant K159R induced expression of FLAG-tagged eIF4E protein in LNCaP paclitaxel resistance, increasing cell viability (Figure 5d) cells (Figure 3c) confers resistance to combine treatment and decreasing apoptotic rates (Figure 5e) when with paclitaxel and androgen withdrawal (Figure 3d) compared with FLAG-tagged eIF4E wt. These results using clonogenic survival test. Overexpression of eIF4E suggest that decreased eIF4E ubiquitination is asso- mimics the effects of Hsp27 on androgen withdrawal ciated with resistance to androgen withdrawal and and paclitaxel resistance on cell viability (Figure 3e) and paclitaxel. apoptosis (Figure 3f), suggesting that eIF4E is an important effector of Hsp27 functions. Protection from proteasomal degradation of eIF4E is independent of 4E-BP1 Hsp27 colocalizes and interacts directly with eIF4E To determine whether ubiquitination of eIF4E results To further define the role of Hsp27 in the regulation of in proteasomal degradation, eIF4E half-life was or interaction with eIF4E, we examined whether Hsp27 determined after OGX-427 treatment, in the presence colocalizes and interacts with eIF4E using immuno- or absence of the proteasome inhibitor MG132 and fluorescence and co-immunoprecipitation. Using con- protein synthesis inhibitor cyclohexamide. Lysates from focal microscopy, Hsp27 colocalized in the cytoplasm LNCaP cells±MG132 (10 mmol/l) and cyclohexamide with eIF4E in LNCaP-Mock and LNCaP-Hsp27 cell (10 mg/ml) were analyzed using western blot with an lines, and the intensity of eIF4E staining and colocaliza- anti-eIF4E antibody. As shown in Figure 6a, MG132 tion with Hsp27 was higher in LNCaP-Hsp27 relative to treatment prolonged the half-life of eIF4E and reversed Mock-transfected cells (Figure 4a). Using co-immuno- the effect of OGX-427-induced Hsp27 knockdown, precipitation with cleared lysate (Figure 4b) and indicating that decreases in eIF4E protein levels recombinant proteins (Figure 4c), we observed that after OGX-427 treatment occurs through proteasomal Hsp27 co-immunoprecipitates with eIF4E in both degradation. Furthermore, as 4E-BP1 has been recently LNCaP-Mock and LNCaP-Hsp27 cells. Total cell lysate described to stabilize and protect eIF4E against confirmed the increase of eIF4E in LNCaP overexpres- ubiquitination and proteasomal degradation in human sing Hsp27. Collectively, the results illustrated in embryonic kidney 293T cells (Murata and Shimotohno, Figures 2, 3 and 4 indicate that Hsp27 interacts directly 2006), we next tested whether Hsp27-mediated inhibition with eIF4E, and that eIF4E levels directly correlate with of eIF4E ubiquitination and proteasomal degradation in that of Hsp27. LNCaP cells required the presence of 4E-BP1–eIF4E binding. LNCaP-Mock or LNCaP-Hsp27 cells were Hsp27 decreases treatment-induced eIF4E ubiquitination treated with 5 nM 4E-BP1 siRNA or control and To elucidate how Hsp27 regulates eIF4E protein harvested for eIF4E using western blot at 72 h after levels, we tested Hsp27 effect on rates of eIF4E transfection. As shown in Figure 6b, 4E-BP1 knock- ubiquitination, and its subsequent proteasomal degra- down does not affect eIF4E levels in LNCaP-Hsp27 dation. LNCaP-Mock and LNCaP-Hsp27 or LNCaP when compared with LNCaP-Mock cells. These data cells treated with OGX-427 or ASO control were suggest that Hsp27-mediated protection against transfected with pcDNA3.1 containing tagged-hemag- eIF4E ubiquitination in PC cells does not require glutinin ubiquitin complementary DNA (cDNA) or the presence of 4E-BP1. Moreover, our data showed pcDNA3.1 empty plasmid control, and ubiquitinated that overexpression of Hsp27 inhibits eIF4E-4E-BP1 levels of eIF4E was determined using co-immunopreci- interaction (Figure 6c), suggesting that both Hsp27 pitation. Figure 5a shows that overexpression of and 4E-BP1 protect eIF4E from ubiquitination and Hsp27 decreases ubiquitinated eIF4E levels, as shown degradation.

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1888 siRNA Concentration control elF4E (nM) siRNA siRNA 251020 2510 20 anti-GAPDH IB 40kDa anti-eIF4E IB 25kDa

Crystal Violet analysis 200 180 control siRNA 20nM 160 eIF4E siRNA 20nM 140 120 100 80 *** 60 *** Cell viability 40

(% age of control) 20 0 LNCaP-Mock LNCaP-Hsp27

empty vector eIF4E-FLAG wt

anti-FLAG IB eIF4E-FLAG

anti-GAPDH IB

Empty vector

eIF4E-FLAG wt

Crystal Violet analysis Flow Cytometry analysis 140 120 *** 120 100 100 80 *** 80 60 60 40 40

20 20

Cell viability (% age of control) 0 0 Sub G0 phase (% age of control) Empty eIF4E-FLAG Empty eIF4E-FLAG vector wt vector wt Figure 3 Cytoprotection induced by Hsp27 is in part mediated by eIF4E. (a) LNCaP cells were treated with indicated concentrations of eIF4E- or control-siRNA for 1 day. At 3 days after transfection, proteins were extracted and analyzed using western blot. (b) LNCaP-Mock and LNCaP-Hsp27 cells were treated with 20 nM of eIF4E- or control-siRNA. After 2 days, serum-free media containing paclitaxel (10 nM) was added for 24 h and cells were analyzed for growth rates using crystal violet assay. The experiment was repeated in triplicate. *** differs from LNCaP-mock control (Pp0.001) using Student’s t-test. (c) The full-length cDNA for FLAG-tagged eIF4E wt was subcloned into the pcDNA3.1 and transfected into LNCaP. Proteins were extracted from cultured cells and FLAG-tagged eIF4E wt and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) protein levels were analyzed using western blotting. (d) Clonogenic survival experiment was performed in LNCaP-eIF4E and LNCaP-empty cells treated for 24 h with paclitaxel (10 nM) in serum-free media. Cell viability was determined using crystal violet (e) and apoptotic rates using ﬂow cytometry (f). The experiment was repeated in triplicate. *** differs from empty vector control (Pp0.001) using the Student’s t-test.

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1889

LNCaP-Mock LNCaP-Hsp27 LNCaP-MockLNCaP-Mock rPI IP LNCaP-Hsp27 rHsp27LNCaP-Hsp27 IP rPI IP rHsp27 IP

anti-eIF4E IB 25kDa Hsp27 IF anti-Hsp27 IB 27kDa

Hsp27 IP

LNCaP-Mock LNCaP-Hsp27

LNCaP-MockLNCaP-MockLNCaP-Hsp27LNCaP-Hsp27

eIF4E IF anti-eIF4E IB 25kDa

anti-GAPDH IB 40kDa

TCL

LNCaP-Mock LNCaP-Hsp27

rPI IP rHsp27 IP

merged anti-eIF4E IB

Hsp27 IP

anti-Hsp27 IB

Hsp27 IP Figure 4 Hsp27 directly interacts with eIF4E. (a) LNCaP-Mock and LNCaP-Hsp27 cells were grown on glass coverslip with RPMI (5% fetal bovine serum) for 48 h. Cells were then fixed and incubated with mouse monoclonal Hsp27 (red immunofluorescence, IF) and rabbit polyclonal eIF4E (green immunofluorescence, IF) antibody. Analysis of focal colocalization was performed using Image Proplus 6 software (MediaCybernetics) with an assignment of yellow for a colocalized foci (merged) and green or red as non- colocalization. (b) LNCaP-Mock and LNCaP-Hsp27 cells were harvested and cell lysate used to immunoprecipitate (IP) Hsp27 using rabbit anti-Hsp27 antibodies (rHsp27) or rabbit pre-immune serum as control (rPI). After resolving immunocomplexes on 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and transfer to polyvinylidene difluoride membrane, the membrane was immunoblotted (IB) with anti-eIF4E antibody. Total cell lysate (TCL) represents proteins from LNCaP-Mock versus LNCaP-Hsp27 cells, extracted from cultured cells and blotted as control with eIF4E and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody. (c) eIF4E and Hsp27 recombinant proteins were used for IP. Hsp27 was IP using rabbit anti-Hsp27 antibodies (rHsp27) or rabbit pre-immune serum as control (rPI). After resolving immunocomplexes on 10% SDS–PAGE and transfer to polyvinylidene difluoride membrane, the membrane was IB with anti-eIF4E antibody and anti-Hsp27 antibody as control.

Discussion exert an effect at multiple control points across various apoptotic pathways to ensure that stress-induced Androgen ablation induces apoptotic PC cell death in damage does not inappropriately trigger cell death 70–80% of men with advanced PC. Unfortunately, (Garrido et al., 2006). Several mechanisms account remissions are temporary and after approximately 18–24 for the cytoprotective effect of Hsp27, including: months of androgen blockade, most patients progress to (1) chaperone inhibitor of protein misfolding, (2) CRPC (Lucas and Petrylak, 2006). Recently, docetaxel- inhibition of key effectors of the apoptotic machinery based chemotherapy has prolonged the median overall at the pre- and post-mitochondrial level (Joza et al., survival by B2–3 months (Gallagher and Gapstur, 2006). 2001; Garrido et al., 2003; Rocchi et al., 2005, 2006) and Cellular and molecular events that mediate CRPC (3) proteasome-mediated degradation of proteins under progression are complex and multifactorial (Rocchi stress conditions (Garrido et al., 2006). Targeting Hsp27 et al., 2004a) and lead to changes in the expression of by the second-generation ASO, OGX-427, inhibited various apoptosis-regulating genes, including androgen Hsp27 expression and enhanced drug sensitivity in receptor, bcl-2 (B-cell CLL/lymphoma 2), clusterin and several xenograft models (Rocchi et al., 2004b, 2005; insulin-like growth factor binding proteins 2 and 5 (Gleave Kamada et al., 2007). A phase I/II clinical trial using et al., 1999; Miyake et al., 2000a, b; Kiyama et al., 2003). OGX-427 in patients with prostate and other cancers is Hsp27 expression is induced by various stressors, such now underway in the United States and Canada (Hotte as chemotherapy and androgen withdrawal, and can et al., 2009; http://www.oncogenex.ca/).

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1890

ock ock ol 70 nM LNCaP-M LNCaP-Hsp27 LNCaP-M LNCaP-Hsp27 empty empty Ub Ub vector vector empty empty ASO ContrOGX-427 70nM 250kDa Ub Ub vector vector 250kDa 250kDa

Ub-eIF4E anti-Ub IB anti-Ub IB Ub-eIF4E anti-HA IB Ub-HA

25kDa 25kDa 20kDa anti-eIF4E IB 25kDa anti-eIF4E IB 25kDa TCL eIF4E IP eIF4E IP

empty vectoreIF4E-FLAG wt eIF4E-FLAG K159R

anti-FLAG IB eIF4E-FLAG

anti-GAPDH IB

Crystal violet analysis Flow Cytometry analysis 160 120 *** 140 100 120 80 100

80 60 *** 60 40 40 20 20 Cell viability (% age of control)

0 Sub G0 phase (% age of control) 0 Empty eIF4E-FLAG eIF4E-FLAG Empty eIF4E-FLAG eIF4E-FLAG vector wt K159R vector wt K159R Figure 5 Hsp27 decreases eIF4E ubiquitination and induces androgen withdrawal and chemo-resistance. (a) LNCaP-Mock and LNCaP-Hsp27 cells were transfected with a pcDNA3.1 plasmid containing ubiquitin (Ub) hemagglutinin (HA)-tagged or pcDNA3.1 empty vector as control. After 48 h, cells were harvested, and lysate used to immunoprecipitate (IP) eIF4E using anti-eIF4E antibody, followed by immunoblotting (IB) with an anti-ubiquitin antibody. Total cell lysate (TCL) represents proteins from LNCaP-Mock versus LNCaP-Hsp27 cells transfected with pcDNA3.1 plasmid containing ubiquitin HA-tagged or pcDNA3.1 empty vector control, extracted from culture cells and blotted as control with HA antibody. (b) LNCaP cells were treated with Hsp27 ASO OGX-427 or ASO control and lysate used to IP eIF4E using anti-eIF4E antibody. After resolving immunocomplexes on 10% sodium dodecyl sulfate– polyacrylamide gel electrophoresis (SDS–PAGE) and transfer to polyvinylidene diﬂuoride membrane, the membrane was IB with anti- eIF4E antibody or anti-ubiquitin antibody. (c) The full-length cDNA for FLAG-tagged eIF4Ewt, ubiquitination site mutation (K159R) was subcloned into the pcDNA3.1 and transfected transiently into LNCaP. After 48 h, proteins were extracted from cultured cells and FLAG-tagged eIF4E wt, K159R and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) protein levels were analyzed using western blotting. At 2 days after transfection, cells were treated with paclitaxel (10 nM) in serum-free media for 24 h, and cell viability was determined using crystal violet (d) and apoptotic rates using ﬂow cytometry (e). The experiment was repeated in triplicate. *** differs from empty vector control (Pp0.001) using the Student’s t-test.

Although mechanisms by which Hsp27 inhibits involved in PC progression suggested a correlation apoptosis are partially deﬁned, its role in AI growth between Hsp27 and the protein translational initiation remains less clear. Only recent studies report its process (not shown). Furthermore, microarray analysis association with hormone-refractory disease (Bubendorf shows that Hsp27 knockdown using OGX-427 corre- et al., 1999; Rocchi et al., 2004a, 2005). In this study we lated with a signiﬁcant decrease in the expression of show that overexpression of Hsp27 in human prostate eIF4E binding protein, 4E-BP1. 4E-BP1 is activated by androgen-dependent LNCaP provides resistance to mammalian target of rapamycin that has a critical role apoptosis induced by androgen withdrawal and pacli- in controlling protein synthesis, and hence, in cell taxel chemotherapy. Interestingly, two-hybrid experi- survival and proliferation through its interaction with ments and western blots performed on several proteins eIF4E (Heesom et al., 2001; Topisirovic et al., 2004).

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1891 This initial observation led us to analyze the inter- cell growth or oncogenesis (Mamane et al., 2004), relationships between Hsp27 and the mammalian target reﬂecting its critical role in protein synthesis (Murata of rapamycin translation initiation pathway. We found and Shimotohno, 2006). eIF4E has been recently that Hsp27 levels directly correlated with eIF4E, which described to regulate cell survival of many human functions to bind mRNA during an early step in the cancers and it is now recognized as a therapeutic target initiation of protein synthesis and facilitates ribosome in several cancers (Mamane et al., 2004; Averous and binding by unwinding mRNA secondary structure Proud, 2006; Petroulakis et al., 2006; Graff et al., 2008), (Richter and Sonenberg, 2005). Considerable evidence including advanced PC (Graff et al., 2009). In spite of suggests that overproduction of eIF4E leads to aberrant the importance of eIF4E, little is known about the role of eIF4E in CRPC progression and the factors regulating eIF4E protein levels. In this report, we identify a Control mechanism by which changes in Hsp27 expression serves ASO OGX-427 -MG132 (hours) as an upstream regulator of eIF4E. 0 24 48 72 0 24 48 72 Transcription of eIF4E gene is induced in response to many stimuli, such as serum and growth factors anti-eIF4E IB 25kDa (Rosenwald et al., 1993). However, mechanisms regulating eIF4E degradation remain incompletely deﬁned. anti-Hsp27 IB 27kDa Othumpangat et al. (2005) recently reported that eIF4E is ubiquitinated and degraded in a proteasome-depen- anti-GAPDH IB 40kDa dent manner. Ubiquitin is a low-molecular-weight polypeptide covalently conjugated to Lys residues in Control target proteins that serves as signal for delivery to and ASO OGX-427 proteolysis by the proteasome (Pickart, 2001). The +MG132 (hours) ubiquitin–proteasome pathway is an important factor 0 24 48 72 0 24 48 72 controlling the expression and activity of regulatory anti-eIF4E IB 25kDa proteins, such as transcription factors, cell cycle regulators and signal transduction proteins (Mitchell, anti-Hsp27 IB 27kDa 2003). Recently, Arora et al. (2005) showed that the ubiquitin–proteasome pathway regulates the turnover of protein translation and cell proliferation through 40kDa anti-GAPDH IB ubiquitination and proteasomal degradation of the eukaryotic elongation factor-2 kinase. They reported that eukaryotic elongation factor-2 kinase complexed with the molecular chaperone Hsp90, which maintained the level of the properly folded kinase. Disruption of this complex using Hsp90 inhibitor, geldanamycin, causes eukaryotic elongation factor-2 kinase ubiquitination control siRNA4E-BP1 5 nM siRNAcontrol 5 nMsiRNA4E-BP1 5 nM siRNA 5 nM and proteasomal degradation (Arora et al., 2005). LNCaP-Mock LNCaP-Hsp27 Our study shows a new mechanism by which a chaperone protein regulates both the mRNA transla- anti-4E-BP1 IB 15-20kDa tional process and resistance to stress in LNCaP cells. Hsp27 colocalizes and interacts directly with eIF4E, and eIF4E protein levels correlate directly with those of anti-eIF4E IB 25kDa

anti-GAPDH IB 40kDa Figure 6 Protection from proteasomal degradation of eIF4E is independent of 4E-BP1. (a) LNCaP cells were treated with Hsp27 ASO OGX-427 or ASO control for 2 days. After transfection, cells were harvested for protein extraction or pretreated with cycloheximide (10 mg/ml) followed by MG132 (10 mmol/l) for indicated time periods. Cell extracts were prepared and run on 10% sodium LNCaP-Mock LNCaP-Hsp27 dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE), followed by western blot analysis using an anti-eIF4E, anti-Hsp27 and anti- glyceraldehyde-3-phosphate dehydrogenase (GAPDH) anti-eIF4E IB antibody. (b) LNCaP-Mock and LNCaP-Hsp27 cells were treated with 5 nM of 4E-BP1 siRNA for 1 day. Proteins were extracted 3 days after the ﬁrst treatment, and 4E-BP1, eIF4E and GAPDH were analyzed using immunoblotting (IB). (c) LNCaP-Mock and anti-4EBP1 IB LNCaP-Hsp27 cells were harvested and cell lysate used to immunoprecipitate (IP) 4E-BP1 using rabbit anti-4E-BP1 antibodies. After resolving immunocomplexes on 10% SDS–PAGE 4E-BP1 IP and transfer to polyvinylidene diﬂuoride membrane, the membrane was IB with anti-eIF4E or anti-4E-BP1 antibody.

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1892 cellular stress: androgen withdrawal- and chemo-therapy

ub ub ub

ub ub Protein ub Protein Hsp27 OGX-427 Synthesis eIF4E Synthesis and eIF4E Inhibition and Cell survival Cell death

19S 19S 20S proteasome 20S

19S 19S

Figure 7 Schematic model illustrating the role of Hsp27 in preventing ubiquitination and degradation of eIF4E during treatment stress induced by androgen withdrawal and chemotherapy.

Hsp27. Accordingly, the half-life of eIF4E is markedly overexpression (Figures 3e and f), suggesting that eIF4E decreased in LNCaP cells treated with Hsp27 inhibitor, could be a major effector of Hsp27 function. As Hsp27 OGX-427 (Figure 2b). The decrease in eIF4E content is has attracted attention as a new therapeutic target for associated with an increase in its ubiquitinated species cancer, targeting Hsp27–eIF4E interaction may provide (Figure 5b) and can be reversed by the proteasome a potent and selective new anticancer agent in humans. inhibitor MG-132 (Figure 6a). We conclude that OGX- Recently, a similar approach has been developed for 427-mediated Hsp27 knockdown reduces eIF4E stabi- shepherdin and AICAR, the two inhibitors of interac- lity, enhancing its ubiquitination and degradation, tions between Hsp90–survivin, creating an opportunity thereby reducing cell viability after androgen with- to selectively induce antitumor activity without any drawal and/or chemotherapy. As a stress-induced effect on the viability of normal cells or tissues (Plescia molecular chaperone, increases in Hsp27 associated et al., 2005; Meli et al., 2006). with androgen withdrawal or chemotherapy stabilize In summary, we used Hsp27 gain- and loss-of- eIF4E by maintaining its properly folded state, and function approaches in vitro and in vivo to highlight block its Lys 159 ubiquitination and proteasome novel mechanisms mediating Hsp27 cytoprotection in degradation. Lys 159 mutation, K159R, has been found LNCaP cells after androgen withdrawal and chemother- to block eIF4E ubiquitination (Murata and Shimotoh- apy. Increased Hsp27 associated with treatment stress no, 2006) and to induce treatment resistance (Figure 5). stabilizes eIF4E by inhibiting its stress-induced ubiqui- This mechanism, which does not require eIF4E binding tination and proteasomal degradation, thus identifying protein 4E-BP1 in our model, leads to protein synthesis another mechanism by which Hsp27 enhances cell and cell survival (Figure 7). Moreover, our data showed survival. Targeting Hsp27–eIF4E interaction may serve that overexpression of Hsp27 inhibits eIF4E-4E-BP1 as a potential therapeutic target in advanced PC. interaction (Figure 6c), suggesting that both Hsp27 and 4E-BP1 protect eIF4E from ubiquitination and degradation. In further support of positive interactions between Materials and methods Hsp27 cytoprotection and eIF4E activity, eIF4E silen- Tumor cell lines cing using siRNA suppressed cytoprotection normally The human prostatic cancer cell line LNCaP was kindly afforded by high Hsp27 levels (Figures 3a and b), provided by Dr Leland WK Chung (Emery University, suggesting that eIF4E is, in part, responsible for Hsp27 Atlanta, GA, USA) and maintained in RPMI-1640 (Life anti-apoptotic activity. Conversely, overexpression of Technologies, Inc., Gaithersburg, MD, USA) supplemented eIF4E in LNCaP cells mimics the effects of Hsp27 with 5% fetal calf serum.

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1893 Lentiviral infection of Hsp27 into LNCaP cells Flow cytometric analysis The full-length cDNA for human Hsp27 was sub-cloned into Flow cytometry of propidium iodide-stained nuclei was the lentiviral vector pHR’-CMV-EGFP at the BamHI and performed as described previously (Rocchi et al., 2004b). XhoI sites as described before (Rocchi et al., 2005). In brief, cells were plated at the density of 106 cells into 10 cm dishes in RPMI-1640 supplemented with 5% fetal calf serum. To assess the effect of Hsp27 overexpression on apoptosis Stable transfection of eIF4E into LNCaP cells induced by androgen withdrawal and paclitaxel, LNCaP- LNCaP cells were stably transfected the day after seeding with Mock and LNCaP-Hsp27 cells were treated the day after pcDNA3.1 vector containing FLAG-tagged eIF4E wt seeding with different concentrations of paclitaxel in serum- (LNCaP-eIF4E) or empty vector control (LNCaP-empty) free media. To assess the effect of eIF4E wt and K159R cells. Cells were then treated with G418 (1 mg/ml) for 2 weeks overexpression, LNCaP cells were transiently transfected with to select one single stable colony from each transfection pcDNA3.1 vector containing FLAG-tagged eIF4Ewt, K159R LNCaP-eIF4E and LNCaP-empty. Each colony was then or empty vector control for 48 h. Cells were then treated in amplified for 6 weeks under G418 treatment and cells were RPMI-1640 serum-free media with 10 nM of paclitaxel. After placed in six-well plates before being treated with paclitaxel 24 h, the cells were analyzed for relative DNA content on a (10 nM) in serum-free media (mimics androgen-withdrawal dual laser flow cytometer (Beckman Coulter Epics Elite, in vitro) for 24 h. Beckman Inc., Miami, FL, USA). Each assay was performed in triplicate. Chemotherapeutic and chemical agents Assessment of in vivo tumor growth Paclitaxel was purchased from Biolyse Pharma (St Catherines, 6 Ontario, Canada). Stock solutions of paclitaxel were prepared For in vivo study, 10 LNCaP-Mock and LNCaP-Hsp27 cells with phosphate-buffered saline (PBS) to the required concen- were inoculated subcutaneously with 0.1 ml of Matrigel trations before each in vitro experiment. Dr Helen M Burt (Becton Dickinson Labware, Franklin Lakes, NJ, USA) in (Pharmaceutical Sciences, University of British Columbia, the flank region of 6- to 8-week-old male athymic nude mice Canada) generously supplied polymeric micellar paclitaxel (Harlan Sprague Dawley, Inc., Indianapolis, IN, USA) used for in vivo studies. Cycloheximide and MG-132 through 27-gauge needle under halothane anesthesia. All were purchased from Calbiochem (Merck Chemicals Ltd, animal procedures were performed according to local guide- Nottingham, UK). lines on animal care and with appropriate institutional certification. Tumors were measured twice weekly and their volumes were calculated by the formula length Â width Â depth Â 0.5236. Mice bearing tumors between 300 and Clonogenic survival experiment 400 mm3 in volume were castrated through scrotal approach To assess the effect of eIF4E overexpression on apoptosis and a total of 0.5 mg micellar paclitaxel was administered induced by androgen withdrawal and paclitaxel, LNCaP cells intravenously three times per week from days 7 to 14 and from were stably transfected with pcDNA3.1 vector containing days 21 to 28. Tumor volume and serum PSA measurements FLAG-tagged eIF4E wt, or empty vector control for 48 h. were performed weekly. Data points for both sets of Cells were then treated with 10 nM of paclitaxel in serum-free experiments were expressed as average tumor volume±s.e. media, and crystal violet assays were performed after 24 h of based on 10 determinations. treatment. Assay was performed in triplicate.

Immunostaining for in situ apoptosis and proliferation index 2 In vitro mitogenic assay LNCaP-Mock and LNCaP-Hsp27 cells were plated in 9.87 cm The in vitro effects of Hsp27 and eIF4E overexpression and labtek (Nunc, Rosklide, Denmark). After 1 day, cells were downregulation using siRNA on apoptosis induced by treated with paclitaxel 10 nM in serum-free media for 24 h paclitaxel and androgen withdrawal were assessed using the before being harvested and fixed with methanol for 10 min. crystal violet assay as previously described (Gleave et al., Slides were then transferred into coplin jar containing 50 ml of 1999). In brief, cells were seeded in each well of 12-well 50% formamide (v/v distilled water) in water bath at 56–60 1C microtiter plates (50 000 cells per well) and allowed to attach for 20 min. Slides were then incubated for 10 min in 0.2% overnight. To study the effect of Hsp27 overexpression on Triton. After washing three times with PBS, endogenous apoptosis induced by androgen withdrawal and paclitaxel, peroxidase was quenched in 3% hydrogen peroxide for 5 min. LNCaP-Mock and LNCaP-Hsp27 cells were treated the day Detection of cleaved apoptotic DNA fragments were after seeding with different concentrations of paclitaxel in performed using mouse monoclonal antibody MAB3299 serum-free media (mimics androgen withdrawal in vitro), and (Chemicon International, Temecula, CA, USA) that specifi- crystal violet assay was performed after 24 h. To assess the cally reacts with single-stranded DNA, permitting detection of effect of eIF4E downregulation on Hsp27 cytoprotection, apoptotic cells and not necrotic cells (Frankfurt et al., 1997). LNCaP-Mock and LNCaP-Hsp27 cells were transiently Proliferation index was detected using a rabbit monoclonal transfected the day after seeding with 20 nM of eIF4E siRNA antibody specific for Ki67 nuclear protein (Thermo Scientific, for 48 h. Cells were then treated with 10 nM of paclitaxel in Fremont, CA, USA). LSAB þ detection kit from DAKO Kit serum-free media and, crystal violet assays were performed the (Dako, Carpinteria, CA, USA) was used as the detection day after. To assess the effect of eIF4E overexpression on system. The Chromogen Nova-Red (Vector Laboratories, apoptosis induced by androgen withdrawal and paclitaxel, Burlingame, CA, USA) was applied for 1 min and counter- LNCaP cells were transiently transfected the day after seeding staining was performed with hematoxylin (Vector Labora- with pcDNA3.1 vector containing FLAG-tagged eIF4E wt, tories, Burlingame, CA, USA). Lysine 159 mutated (K159R) or empty vector control for 48 h. Cells were then treated with 10 nM of paclitaxel in serum-free Scoring of single-strand DNA and ki67 staining media and, crystal violet assays were performed after 24 h of The staining intensity of labeled cells was evaluated and scored treatment. Each assay was performed in triplicate. by one pathologist (LF), and automated quantitative image

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1894 analysis by pro-plus image software (MediaCybernetics, San 5 Â SSC, 0.01% sodium dodecyl sulfate, 8 mgbovineserum Diego, CA, USA) was used. The overall percentage of cells albumin, 25 mg yeast tRNA and 20 mgsalmontestesDNA.After showing staining (0–100%) was indicated. stringent washes (1 Â SSC and 0.1% sodium dodecyl sulfate, andthen0.1Â SSC), fluorescent images of the slides were acquired using an ArrayWoRx, Microarray Scanner (Applied Statistical analysis Precision, Issaquah, WA, USA). Signal quality and quantity All the results were expressed as mean±s.e. Statistical analysis were assessed using Imagene 5.6 (BioDiscovery, San Diego, CA, was performed using one-way analysis of variance followed by USA). Data from Imagene were analyzed in GeneSpring 6.1 Fisher’s protected least significant difference test (Statview using per-spot and per-chip intensity-dependent (LOWESS) 512, Brain Power Inc., Calabases, CA, USA). *P 0.05 was p normalization (Silicon Genetics, Mississauga, ON, Canada) for considered significant, with **P 0.01 and ***P 0.001. p p profiling significant changes in gene expression.

Antisense oligodeoxynucleotide and siRNA sequences Hsp27 ASO sequences were manufactured by ISIS Pharma- Western blot analysis ceuticals (Carlsbad, CA, USA) and supplied by OncoGenex Western Blot analysis was performed as described previously Technologies (Vancouver, British Columbia, Canada). The (Rocchi et al., 2004b) with 1:5000 rabbit anti-Hsp27 monoclonal ASOs were second generation with a 20-O-(2-methoxy) ethyl antibody (Stressgen Bioreagents, Victoria, Canada), 1:1000 rabbit modiﬁcation. Hsp27 ASO sequence corresponding to the anti-4E-BP1 polyclonal antibody (Cell Signaling Technology, human Hsp27 translation initiation site was 50-GGGACGCG Inc., Danvers, MA, USA), 1:1000 rabbit anti-eIF4E polyclonal GCGCTCGGTCAT-30 and designated OGX-427 (OncoGenex antibody (Cell Signaling Technology), 1:500 mouse anti-ubiquitin Technologies), whereas the scramble control sequence was monoclonal antibody (Santa Cruz Biotechnology, Inc., Heidelberg, 50-CAGCGCTGACAACAGTTTCAT-30 and designated ASO Germany), 1:1000 mouse monoclonal FLAG antibody (Sigma, control. The 4E-BP1 and eIF4E siRNA was the Hs_eI- St Louis, MO, USA) and 1:1000 mouse monoclonal HA antibody F4EBP1_2_HP and the Hs_eIF4E_1_HP validated siRNA (Roche Applied Science, Meylan, France). Loading levels were from Qiagen (Courtboeuf, France). The control siRNA was normalized using 1:5000 rabbit anti-glyceraldehyde-3-phosphate also from Qiagen. dehydrogenase polyclonal antibody (Abcam Inc., Cambridge, MA, USA).

Treatment of cells with ASO and siRNA Cells were plated at a density of 25 000 cells by 1.9 cm2 and Ubiquitin and eIF4E transfection treated after 1 day with indicated siRNA or ASO for 1 or 2 Hemagglutinin-tagged ubiquitin in pcDNA3.1 have been days, respectively. Oligofectamine, a cationic lipid (Invitrogen, described previously (Soubeyran et al., 2002). LNCaP-Mock Life Technologies, Burlington, ON, Canada), was used and LNCaP-Hsp27 or LNCaP cells treated with OGX-427 or to increase siRNA or ASO uptake into the cells. Cells were ASO control were transfected with pcDNA3.1 vector containing treated with indicated siRNA or ASO concentrations after a hemagglutinin–ubiquitin or pcDNA3.1 empty vector control preincubation for 20 min with 3 mg/ml oligofectamine in serum- using lipofectamine reagent (Invitrogen, Cergy Pontoise, France) free OPTI-MEM (Life Technologies, Inc.). After 4 h, the medium following the manufacturer’s instructions. A total of 5 mgof was replaced with standard culture medium described above. cDNA was used when cells were transfected in 10 cm dishes. To study the effects on rates of eIF4E proteasome degradation, Cells were lysed in lysis buffer (48 h after transfection) and cyclohexamide (10 mg/ml) and MG132 (10 mmol/l) were added centrifuged for 15 min at 4 1C. Cleared lysates were used for at the end of the second transfection with ASO in replaced immunoprecipitation and western blot analyses. FLAG-tagged medium for 12, 24, 48 and 72 h. eIF4E wt or ubiquination mutant K159R (Murata and Shimotohno, 2006) was kindly provided by Dr Takayuki Murata (McGill University, Montreal, Canada). LNCaP cells Microarray experiments were transfected with pcDNA3.1 vector containing FLAG- Total RNA from LNCaP-Mock and LNCaP-Hsp27 and LNCaP tagged eIF4E wt, K159R or empty vector control using the same cells treated with OGX-427 or ASO control was compared on the protocol that was previously described for ubiquitin transfection. same chip. A dye-swap for each pair was performed to account for dye bias. Three samples from each point were analyzed and results were confirmed using northern analysis (not shown). Microarrays Immunofluorescence of 13 791 (70-mer) human oligos (Operon) printed in duplicate LNCaP-Mock and LNCaP-Hsp27 cells were grown on glass in 3 Â saline sodium citrate (SSC) into amine-coated slides coverslips in RPMI media and 5% fetal bovine serum for 48 h. (Ezray, Apogent, Fisher Scientific, Ottawa, ON, Canada) were Subsequently, cells were fixed with cold 4% paraformaldehyde supplied by the Array Facility of the Prostate Centre at for 10 min at À20 1C and permeabilized in 0.05% Triton X-100 Vancouver General Hospital. Slides were printed with a in PBS. Slides were incubated in blocking solution of 1% BioRobotics Microgrid II (Harvard Biosciences and Genomic bovine serum albumin in PBS for 30 min and treated with Solutions, Ann Arbor, MI, USA) at 60% humidity at 22 1C primary antibodies mouse monoclonal Hsp27 (StressGen) and and were then ultraviolet crosslinked using 3000 mJ Stratalinker rabbit polyclonal eIF4E (Cell Signaling Technology) simulta- (Stratagene, La Jolla, CA, USA). Slides were pre-hybridized in neously overnight. Secondary fluorescent antibodies goat anti- 5 Â SSC, 0.1% sodium dodecyl sulfate and 0.2% bovine serum mouse Alexa Fluor 546 (Invitrogen) and goat anti-rabbit albumin (Sigma Chemical Co., St Louis, MO, USA) at 48 1Cfor Alexa Fluor 488 (Invitrogen) were added for 1 h at room 45 min, washed in deionized water, dipped in isopropanol and temperature with 3 Â 5 min washes (in PBS). Cells examined dried in a centrifuge at 2000 r.p.m. for 2 min. Arrays were for localization of red and green protein were mounted with hybridized in a humid Hybaid oven (ThermoHybaid, Ashford, fluorescent 4,6-diamidino-2-phenylindole vectashield mounting UK) with reverse-transcribed fluorescently labeled (Cy3- or Cy5- media (Vector Laboratories). Images were captured using a deoxyuridine-triphosphatase) (Amersham-Pharmacia, Arlington Zeiss 510META fluorescence confocal microscope plan 63X/1.4 Heights, IL, USA) cDNA (from 20 mg of total RNA) at 42 1C (Le Pecq, France) followed by analysis of focal colocalization for 16 h in a hybridization buffer consisting of 50% formamide, performed with Image Proplus 6 software (MediaCybernetics,

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1895 Wokingham, UK) with an assignment of yellow for colocalized Conﬂict of interest foci and green or red as non-colocalization. The authors declare no conﬂict of interest.

Immunoprecipitation Cleared lysates with adjusted protein concentration (Protein Acknowledgements assay, Bio-Rad, Marnes-la-Coquette, France) or recombinant Hsp27 (Stressgen) and glutathione S-transferase-tagged eIF4E We thank Dr Valentina Evdokimova (British Columbia proteins (Abnova Corporation, Taipei, Taiwan) were used for Cancer Agency, Canada) for her helpful advice on eIF4E immunoprecipitation with rabbit anti-Hsp27 antibody (Stressgen), experiments. We thank Dr Takayuki Murata (McGill Uni- rabbit anti-eIF4E antibody (Cell Signaling), rabbit anti-FLAG versity, Canada) for providing the pcDNA3.1 FLAG-tagged antibody (Rockland Immunochemicals, Inc., Gilbertsville, PA, eIF4E vectors. We thank Virginia Yago and Paul Jugpal USA), rabbit 4E-BP1 (Cell Signaling) or pre-immune serum (Prostate Centre, Canada) for their excellent technical O/N at 4 1C. Immune complexes were precipitated after 2-h assistance in animal and laboratory experimentation. We incubation with 30 ml of protein A sepharose beads (Zymed thank Dr Jonathan Nowak (Inserm U624, France) for his Laboratories Inc., San Francisco, CA, USA). After washing excellent help in confocal microscopy. This work was three times in cold lysis buffer, the complexes were resus- supported from grants by l’Institut National de la Sante´et pended in protein sample buffer (Bio-Rad) and boiled de la Recherche Me´dicale (INSERM), l’Association pour la for 5 min before western blot was performed as described Recherche sur le Cancer (ARC) and l’Association pour la before. Recherche sur les Tumeurs de la Prostate (ARTP).

References

Armengol G, Rojo F, Castellvi J, Iglesias C, Cuatrecasas M, Pons B in advanced human prostate cancers and significantly related to et al. (2007). 4E-binding protein 1: a key molecular ‘funnel factor’ in reduced patient survival. Cancer Res 69: 3866–3873. human cancer with clinical implications. Cancer Res 67: 7551–7555. Heesom KJ, Gampel A, Mellor H, Denton RM. (2001). Cell cycle- Arora S, Yang JM, Hait WN. (2005). Identification of the ubiquitin- dependent phosphorylation of the translational repressor eIF-4E proteasome pathway in the regulation of the stability of eukaryotic binding protein-1 (4E-BP1). Curr Biol 11: 1374–1379. elongation factor-2 kinase. Cancer Res 65: 3806–3810. Hotte SJ, Yu EY, Hirte HW, Higano CS, Gleave M, Chi KN. (2009). Averous J, Proud CG. (2006). When translation meets transformation: OGX-427, a 20 methoxyethyl antisense oligonucleotide (ASO), the mTOR story. Oncogene 25: 6423–6435. against Hsp27: Results of a first-in-human trial. J Clin Oncol 27: Bubendorf L, Kolmer M, Kononen J, Koivisto P, Mousses S, Chen Y 15s (suppl; abstr 3506). et al. (1999). Hormone therapy failure in human prostate cancer: Joza N, Susin SA, Daugas E, Stanford WL, Cho SK, Li CY et al. analysis by complementary DNA and tissue microarrays. J Natl (2001). Essential role of the mitochondrial apoptosis-inducing factor Cancer Inst 91: 1758–1764. in programmed cell death. Nature 410: 549–554. Cornford PA, Dodson AR, Parsons KF, Desmond AD, Woolfenden Kamada M, So A, Muramaki M, Rocchi P, Beraldi E, Gleave M. A, Fordham M et al. (2000). Heat shock protein expression (2007). Hsp27 knockdown using nucleotide-based therapies inhibit independently predicts clinical outcome in prostate cancer. Cancer tumor growth and enhance chemotherapy in human bladder cancer Res 60: 7099–7105. cells. Mol Cancer Ther 6: 299–308. Frankfurt OS, Robb JA, Sugarbaker EV, Villa L. (1997). Apoptosis in Kiyama S, Morrison K, Zellweger T, Akbari M, Cox M, Yu D et al. breast carcinomas detected with monoclonal antibody to single- (2003). Castration-induced increases in insulin-like growth factor- stranded DNA: relation to bcl-2 expression, hormone receptors, and binding protein 2 promotes proliferation of androgen-independent lymph node metastases. Clin Cancer Res 3: 465–471. human prostate LNCaP tumors. Cancer Res 63: 3575–3584. Fusi A, Procopio G, Della Torre S, Ricotta R, Bianchini G, Salvioni R Liang JJ. (2000). Interaction between beta-amyloid and lens alphaB- et al. (2004). Treatment options in hormone-refractory metastatic crystallin. FEBS Lett 484: 98–101. prostate carcinoma. Tumori 90: 535–546. Lucas A, Petrylak DP. (2006). The case for early chemotherapy for the Gallagher E, Gapstur R. (2006). Hormone-refractory prostate cancer: treatment of metastatic disease. J Urol 176: S72–5. a shifting paradigm in treatment. Clin J Oncol Nurs 10: 233–240. Mamane Y, Petroulakis E, Rong L, Yoshida K, Ler LW, Sonenberg Garrido C, Brunet M, Didelot C, Zermati Y, Schmitt E, Kroemer G. N. (2004). eIF4E—from translation to transformation. Oncogene (2006). Heat shock proteins 27 and 70: anti-apoptotic proteins with 23: 3172–3179. tumorigenic properties. Cell Cycle 5: 2592–2601. Meli M, Pennati M, Curto M, Daidone MG, Plescia J, Toba S et al. Garrido C, Schmitt E, Cande C, Vahsen N, Parcellier A, Kroemer G. (2006). Small-molecule targeting of heat shock protein 90 chaperone (2003). HSP27 and HSP70: potentially oncogenic apoptosis function: rational identification of a new anticancer lead. J Med inhibitors. Cell Cycle 2: 579–584. Chem 49: 7721–7730. Gleave ME, Hsieh JT, Wu HC, von Eschenbach AC, Chung LW. Miller H, Poon S, Hibbert B, Rayner K, Chen YX, O’Brien ER. (1992). Serum prostate specific antigen levels in mice bearing human (2005). Modulation of estrogen signaling by the novel interaction of prostate LNCaP tumors are determined by tumor volume and heat shock protein 27, a biomarker for atherosclerosis, and estrogen endocrine and growth factors. Cancer Res 52: 1598–1605. receptor beta: mechanistic insight into the vascular effects of Gleave ME, Miayake H, Goldie J, Nelson C, Tolcher A. (1999). estrogens. Arterioscler Thromb Vasc Biol 25: e10–4. Targeting bcl-2 gene to delay androgen-independent progression Mitchell BS. (2003). The proteasome—an emerging therapeutic target and enhance chemosensitivity in prostate cancer using antisense bcl- in cancer. N Engl J Med 348: 2597–2598. 2 oligodeoxynucleotides. Urology 54: 36–46. Miyake H, Nelson C, Rennie PS, Gleave ME. (2000a). Testosterone- Graff JR, Konicek BW, Carter JH, Marcusson EG. (2008). Targeting repressed prostate message-2 is an antiapoptotic gene involved in the eukaryotic translation initiation factor 4E for cancer therapy. progression to androgen independence in prostate cancer. Cancer Cancer Res 68: 631–634. Res 60: 170–176. Graff JR, Konicek BW, Lynch RL, Dumstorf CA, Dowless MS, Miyake H, Pollak M, Gleave ME. (2000b). Castration-induced McNulty AM et al. (2009). eIF4E activation is commonly elevated up-regulation of insulin-like growth factor binding protein-5

Oncogene Increased Hsp27 induces androgen withdrawal and chemoresistance C Andrieu et al 1896 potentiates insulin-like growth factor-I activity and accelerates cers and activators of transcription 3-mediated suppression of progression to androgen independence in prostate cancer models. apoptosis. Cancer Res 65: 11083–11093. Cancer Res 60: 3058–3064. Rocchi P, Jugpal P, So A, Sinneman S, Ettinger S, Fazli L et al. (2006). Murata T, Shimotohno K. (2006). Ubiquitination and proteasome- Small interference RNA targeting heat-shock protein 27 inhibits the dependent degradation of human eukaryotic translation initiation growth of prostatic cell lines and induces apoptosis via caspase-3 factor 4E. J Biol Chem 281: 20788–20800. activation in vitro. BJU Int 98: 1082–1089. Oridate N, Kim HJ, Xu X, Lotan R. (2005). Growth inhibition of head Rocchi P, Muracciole X, Fina F, Mulholland DJ, Karsenty G, Palmari and neck squamous carcinoma cells by small interfering RNAs J et al. (2004a). Molecular analysis integrating different pathways targeting eIF4E or cyclin D1 alone or combined with cisplatin. associated with androgen-independent progression in LuCaP 23.1 Cancer Biol Ther 4: 318–323. xenograft. Oncogene 23: 9111–9119. Othumpangat S, Kashon M, Joseph P. (2005). Eukaryotic translation Rocchi P, So A, Kojima S, Signaevsky M, Beraldi E, Fazli L et al. initiation factor 4E is a cellular target for toxicity and death due to (2004b). Heat shock protein 27 increases after androgen ablation exposure to cadmium chloride. J Biol Chem 280: 25162–25169. and plays a cytoprotective role in hormone-refractory prostate Parcellier A, Brunet M, Schmitt E, Col E, Didelot C, Hammann A cancer. Cancer Res 64: 6595–6602. et al. (2006). HSP27 favors ubiquitination and proteasomal Rojo F, Najera L, Lirola J, Jimenez J, Guzman M, Sabadell MD et al. degradation of p27Kip1 and helps S-phase re-entry in stressed cells. (2007). 4E-binding protein 1, a cell signaling hallmark in breast FASEB J 20: 1179–1181. cancer that correlates with pathologic grade and prognosis. Clin Parcellier A, Schmitt E, Gurbuxani S, Seigneurin-Berny D, Pance A, Cancer Res 13: 81–89. Chantome A et al. (2003). HSP27 is a ubiquitin-binding protein Rosenwald IB, Lazaris-Karatzas A, Sonenberg N, Schmidt EV. (1993). involved in I-kappaBalpha proteasomal degradation. Mol Cell Biol Elevated levels of cyclin D1 protein in response to increased expression 23: 5790–5802. of eukaryotic initiation factor 4E. MolCellBiol13: 7358–7363. Petroulakis E, Mamane Y, Le Bacquer O, Shahbazian D, Sonenberg Ruggero D, Montanaro L, Ma L, Xu W, Londei P, Cordon-Cardo C N. (2006). mTOR signaling: implications for cancer and anticancer et al. (2004). The translation factor eIF-4E promotes tumor therapy. Br J Cancer 94: 195–199. formation and cooperates with c-Myc in lymphomagenesis. Nat Petrylak DP, Tangen CM, Hussain MH, Lara Jr PN, Jones JA, Taplin Med 10: 484–486. ME et al. (2004). Docetaxel and estramustine compared with Soubeyran P, Kowanetz K, Szymkiewicz I, Langdon WY, Dikic I. mitoxantrone and prednisone for advanced refractory prostate (2002). Cbl-CIN85-endophilin complex mediates ligand-induced cancer. N Engl J Med 351: 1513–1520. downregulation of EGF receptors. Nature 416: 183–187. Pickart CM. (2001). Ubiquitin enters the new millennium. Mol Cell 8: Tannock IF, de Wit R, Berry WR, Horti J, Pluzanska A, Chi KN et al. 499–504. (2004). Docetaxel plus prednisone or mitoxantrone plus prednisone Plescia J, Salz W, Xia F, Pennati M, Zaffaroni N, Daidone MG et al. for advanced prostate cancer. N Engl J Med 351: 1502–1512. (2005). Rational design of shepherdin, a novel anticancer agent. Topisirovic I, Ruiz-Gutierrez M, Borden KL. (2004). Phosphorylation Cancer Cell 7: 457–468. of the eukaryotic translation initiation factor eIF4E contributes to Richter JD, Sonenberg N. (2005). Regulation of cap-dependent its transformation and mRNA transport activities. Cancer Res 64: translation by eIF4E inhibitory proteins. Nature 433: 477–480. 8639–8642. Rocchi P, Beraldi E, Ettinger S, Fazli L, Vessella RL, Nelson C et al. Wendel HG, De Stanchina E, Fridman JS, Malina A, Ray S, Kogan S (2005). Increased Hsp27 after androgen ablation facilitates andro- et al. (2004). Survival signalling by Akt and eIF4E in oncogenesis gen-independent progression in prostate cancer via signal transdu- and cancer therapy. Nature 428: 332–337.

Oncogene