BTG2 Loss and Mir-21 Upregulation Contribute to Prostate Cell Transformation by Inducing Luminal Markers Expression and Epithelial–Mesenchymal Transition

ORIGINAL ARTICLE BTG2 loss and miR-21 upregulation contribute to prostate cell transformation by inducing luminal markers expression and epithelial–mesenchymal transition

V Coppola1,6, M Musumeci1,6, M Patrizii1, A Cannistraci1, A Addario1, M Maugeri-Sacca` 2, M Biffoni1, F Francescangeli1, M Cordenonsi3, S Piccolo3, L Memeo4, A Pagliuca1, G Muto5, A Zeuner1, R De Maria2,6 and D Bonci1,6

Prostate cancer is one of the leading causes of cancer-related death in men. Despite signiﬁcant advances in prostate cancer diagnosis and management, the molecular events involved in the transformation of normal prostate cells into cancer cells have not been fully understood. It is generally accepted that prostate cancer derives from the basal compartment while expressing luminal markers. We investigated whether downregulation of the basal protein B-cell translocation gene 2 (BTG2) is implicated in prostate cancer transformation and progression. Here we show that BTG2 loss can shift normal prostate basal cells towards luminal markers expression, a phenotype also accompanied by the appearance of epithelial–mesenchymal transition (EMT) traits. We also show that the overexpression of microRNA (miR)-21 suppresses BTG2 levels and promotes the acquisition of luminal markers and EMT in prostate cells. Furthermore, by using an innovative lentiviral vector able to compete with endogenous mRNA through the overexpression of the 30-untranslated region of BTG2, we demonstrate that in prostate tumor cells, the levels of luminal and EMT markers can be reduced by derepression of BTG2 from microRNA-mediated control. Finally, we show that the loss of BTG2 expression confers to non-tumorigenic prostate cells ability to grow in an orthotopic murine model, thus demonstrating the central role of BTG2 downregulaton in prostate cancer biology.

Oncogene (2013) 32, 1843–1853; doi:10.1038/onc.2012.194; published online 21 May 2012 Keywords: prostate cancer; BTG2; microRNA; basal and luminal phenotype; EMT

INTRODUCTION their epithelial characteristics, whereas acquiring a mesenchymal- Although much progress has been made in prostate cancer like phenotype associated with the acquisition of migratory and diagnosis and management over the last few decades, prostate invasive traits.12 The basal cell compartment is also a major site of tumors still represent one of the leading causes of cancer-related expression of the B-cell translocation gene 2 (BTG2).13 BTG2 is a death in men.1 The malignant transformation can occur through pan-cell cycle modulator and a major effector of p53-induced progressive acquisition of genetic and epigenetic lesions, coupled proliferation arrest that exerts its biological activity by inhibiting with increasing microenvironmental stimuli. However, early the expression of cyclin D1 and the phosphorylation of Rb.14–16 In molecular events underlying the oncogenic process in the addition, knockdown of BTG2 revealed its cooperation with the prostate remain largely unknown. Prostate acini are composed RAS oncogene in inducing cell transformation.17 The tumor- of a luminal compartment of differentiated cells surrounded by suppressive role of BTG2 is supported by its downregulation in an undifferentiated layer of basal cells. A transit-amplifying several tumors, including clear cell renal carcinoma,18 breast population composed of intermediate progenitors retaining cancer19 and prostate adenocarcinoma.20 both basal and luminal markers mirrors a continuum in the It is well established that post-transcriptional modiﬁcations differentiation program.2 Basal cells have been traditionally modulate gene expression. MicroRNAs (miRs) are small, non- considered the cells of origin of prostate cancer.3 However, how coding, single-stranded RNAs that suppress mRNA translation of cancer cells may both express luminal cell markers4,5 and have speciﬁc target genes.21 A compelling body of evidence indicates functional properties of basal cells, such as unlimited self-renewal that miRs control nearly all basic cell functions22 and that their capacity6 and activation of survival mechanisms, continues to deregulation is deeply implicated in tumor development and remain controversial.7 Unlike prostate cancer cells, the basal layer progression.23–25 miR-21 upregulation occurs in many cancers of normal prostate acini expresses considerable levels of DNp63a, including prostate tumors.26–28 It has also been correlated with an isoform of p63 involved in stemness maintenance and tumor aggressiveness, androgen-dependent and independent differentiation control during prostate organogenesis.8,9 In prostate cancer growth, and metastatic spread.29,30 Here we addition, DNp63 loss has been associated with enhanced demonstrate a novel role of miR-21 in driving prostate cells motility and epithelial–mesenchymal transition (EMT),10,11 a toward acquiring a malignant phenotype via the downregulation trans-differentiation process through which cancer cells lose of BTG2.

1Department of Hematology, Oncology and Molecular Medicine, Istituto Superiore di Sanita`, Rome, Italy; 2Regina Elena Cancer Institute, Rome, Italy; 3Department of Histology, Microbiology and Medical Biotechnologies, University of Padua School of Medicine, Padua, Italy; 4Department of Experimental Oncology, Mediterranean Institute of Oncology, Catania, Italy and 5Department of Urology, S. Giovanni Bosco Hospital, Turin, Italy. Correspondence: Professor R De Maria, Regina Elena Cancer Institute, Rome, Italy or Dr D Bonci, Department of Hematology, Oncology and Molecular Medicine, Istituto Superiore di Sanita`, Rome, Italy. E-mail: [email protected] or [email protected] 6These authors contributed equally to this work. Received 30 June 2011; revised and accepted 5 April 2012; published online 21 May 2012 BTG2 loss and miR-21 upregulation V Coppola et al 1844 RESULTS as a transcriptional regulator, we investigated whether p63 BTG2 is downregulated in prostate tumors and cancer cell lines reduction might result in similar features and potentially mediate We evaluated, by immunohistochemistry, the expression of BTG2 the effects of BTG2 downregulation. We therefore performed RNA in five specimens of prostate tumors and five normal prostate interference of p63 in RWPE-1 cells, observing a strong upregula- tissues. The analysis confirmed that BTG2 expression is confined to tion of luminal markers CK8-18 and AR (Supplementary Figure the basal cell layer of normal tissues (60–70% of cells), with a S2B). Therefore, BTG2 and p63 interference both result in the acquisition of a luminal phenotype. As BTG2 has been reported to prevalent staining of both nuclei and cytoplasm, and a minority of 17,35 either nuclear or cytoplasmatic localization. BTG2 staining was suppress RAS activity, we investigated whether the reduction negative in luminal cells of both normal prostate acini and tumor of p63 in BTG2-interfered cells could be because of the activation tissues (Figure 1a). However, in some cases, BTG2 was detected in of the ERK signaling. The concurrent interference of KRAS was indeed able to rescue p63 levels (Supplementary Figure S2C). p63 luminal cells of acini dispersed in tumor areas, as highlighted by 10 co-staining of BTG2 and cytokeratins 8 and 18 (CK8-18; Figure 1b). loss has also been associated with EMT, the process through To further corroborate the downregulation of BTG2 in prostate which epithelial cells acquire a more elongated, motile phenotype tumors, we evaluated its expression in primary cells isolated from associated with the expression of mesenchymal markers like five prostate tumors. Flow cytometry analysis confirmed BTG2 vimentin and fibronectin, and the loss of the epithelial adhesion downregulation in cancer cells as compared with normal primary molecule E-Cadherin. We verified this mechanism in prostate cells cells (Supplementary Figure S1A). We validated the specificity of by evaluating the levels of vimentin upon interference and our BTG2 staining by BTG2 immunoprecipitation and detection overexpression of p63. In RWPE-1 cells, RNA interference against with a non-commercial anti-BTG2 antibody31 (Supplementary all p63 and DNp63 isoforms resulted in upregulation of vimentin Figure S1B). To investigate the consequences of BTG2 down- (Supplementary Figure S2D). Conversely, in RWPE-2 cells, which regulation in prostate cancer, we evaluated BTG2 levels in our express higher levels of vimentin, the overexpression of DNp63 experimental model, comparing its expression in three established reduced this mesenchymal marker, whereas a TAp63-carrying prostate cell lines representing different phases of prostate plasmid did not (Supplementary Figure S2D). This suggests that tumorigenesis: (i) immortalized non-tumorigenic RWPE-1 cells, (ii) only DNp63 isoforms can control the expression of vimentin. We early tumoral RWPE-2 cells and (iii) metastasis-derived PC3 cells. therefore evaluated whether BTG2 interference might also result Cytofluorimetric analysis revealed a progressive reduction of BTG2 in EMT. Flow cytometry and immunofluorescence analyses expression in this series of cell lines (Figure 1c). Notably, in RWPE-1 showed mesenchymal markers induction in BTG2-interfered cells cells, BTG2 protein is strongly expressed in the nuclei of small (Figure 2d and Supplementary Figure S2E). Similarly, E-Cadherin round cells, while residing in the cytoplasm of larger cells staining underlined a partial disorganization of epithelial junc- (Figure 1d). Given the role that BTG2 has as a transcriptional co- tions, and flow cytometry confirmed a 30% reduction of factor, nuclear localization suggests that its transcriptional E-Cadherin-positive cells (Figure 2e). This phenotype correlated function can be exploited only in a subpopulation of cells.32 with an increase in motility, as shown by migration assay in RWPE-1 is a heterogeneous cell line composed of a progenitor Boyden chambers (Figure 2f). subpopulation, an intermediate population co-expressing basal and luminal cytokeratins, and a more differentiated cell miR-21 overexpression reproduces the effects of BTG2 knockdown 33,34 fraction. p63 protein is expressed by most of the population, We next explored the possible mechanisms responsible for BTG2 with higher levels in the smaller, undifferentiated cells, whereas downregulation in prostate cancer. A posttranscriptional control CK8-18 are mainly expressed by larger, differentiated cells of BTG2 in prostate cancer was hypothesized.20 The established (Figure 1e). As prostate tumor cells are largely characterized by oncogene miR-2136,37 is overexpressed in many cancers, including the expression of luminal markers while progenitor cells have a prostate cancer.26,27,30 A direct regulation of BTG2 protein basal phenotype, RWPE-1 cells represent a valuable model for expression by miR-21 has been recently demonstrated in studying phenotype modification during neoplastic laryngeal carcinoma.38 First, we confirmed the direct binding of transformation. BTG2 30-untranslated region (UTR) by miR-21 with luciferase assay in prostate and non-prostate cells (Figure 3a and Supplementary BTG2 knockdown leads acquisition of a luminal phenotype and Figure S3A), whereas p63 did not result to be a miR-21 target EMT (Supplementary Figure S3A and Supplementary information). The prostate basal cell-restricted expression of BTG2, its absence Afterwards, we transduced RWPE-1 cells with the TWEEN (TW) in the nuclei of more differentiated cells and its loss in prostate lentiviral vector,39 engineered to drive the constitutive expression tumors led us to speculate an involvement of BTG2 in maintaining of miR-21. As compared with TW-infected cells, miR-21 the basal compartment. To test this hypothesis, we performed overexpressing RWPE-1 cells downregulated BTG2 protein transient RNA interference of BTG2 in RWPE-1 cells (Figure 2a and expression by more than 50% (Figure 3b). We then analyzed the Supplementary Figure S2A). As compared with control cells, phenotype of RWPE-1 miR-21 cells and observed comparable RWPE-1 transfected with small interefering RNA (siRNA) BTG2 features with BTG2-interfered cells, such as decrease of p63, showed a change in cell morphology, with an increase of larger induction of CK8-18 and AR (Figure 3c), as well as upregulation of spindle shape cells, as revealed by phalloidin staining of the vimentin and fibronectin (Figure 3d) and reduction of E-Cadherin- cytoskeleton (Figure 2b). Evaluation of marker expression demon- positive cells (Figure 3e). In miR-21-overexpressing cells, the strated an increase of the luminal proteins CK8-18 and androgen transfection of DNp63 could reduce CK8-18 and vimentin levels, receptor (AR), with a parallel decrease in the expression of the confirming the participation of this molecule in the observed typical basal marker p63 (Figure 2c). Given the importance of p63 phenotype (Supplementary Figure S3B). Thus, the ectopic

Figure 1. BTG2 is downregulated in prostate cancer cell lines and primary cells. (a) Representative fields of immunohistochemistry for BTG2 and basal cell cytokeratins in five normal prostates (N1–5) and five prostate tumors (T1–5). The table summarizes the percentage of cells positive for BTG2 and its cellular localization. (b) Immunofluorescence (IF) analysis of BTG2 and CK8-18 on cryostatic sections of normal and tumor prostates. Three normal and tumor prostates were analyzed. (c) Cytofluorimetric (fluorescence-activated cell sorting (FACS)) analysis of BTG2 expression in prostate cell lines. (d) IF analysis of BTG2 expression in RWPE-1 cells, alone or merged with DAPI (40,60-diamidino-2- phenylindole). (e) Phase contrast and IF analysis of p63 and CK8-18 in RWPE-1 cells, alone or merged with DAPI. One staining representative of three FACS or IF experiments is reported.

Oncogene (2013) 1843 – 1853 & 2013 Macmillan Publishers Limited BTG2 loss and miR-21 upregulation V Coppola et al 1845 expression of miR-21 was able to recapitulate the mesenchymal- mesenchymal transition in colon cancer cells,41 we conﬁrmed like changes induced by the knockdown of BTG2. As the well- PDCD4 as miR-21 target and investigated whether its knockdown established miR-21 target PDCD4 40 has been shown to inhibit the by short hairpin RNA (shRNA) could reproduce EMT traits.

& 2013 Macmillan Publishers Limited Oncogene (2013) 1843 – 1853 BTG2 loss and miR-21 upregulation V Coppola et al 1846

Figure 2. BTG2 knockdown leads to acquisition of a luminal phenotype and EMT. (a) Cytofluorimetric analysis of BTG2 expression in cells transfected with siRNA scramble and BTG2-interfered siRNA BTG2 cells. Control represents control Ab-incubated cells (b) Phalloidin-mediated staining of the cytoskeleton upon BTG2 interference, merged with DAPI. (c) Western blot analysis of p63, AR and CK8-18 upon BTG2 interference. Actin or tubulin were used as loading controls. (d) Fluorescence-activated cell sorting analysis of vimentin expression upon BTG2 interference. (e) IF analysis of E-Cadherin upon BTG2 interference. Note that for IF an antibody recognizing the intracellular domain of E-Cadherin was used. On the right, histogram representing the decrease in the percentage of E-Cadherin-positive cells, as quantified by cytofluorimetric analysis with an antibody against an extracellular epitope. (f) Migration assay with Boyden chambers in control and BTG2- interfered cells. A Â 10 representative field is shown on the left. On the right a histogram quantifies relative migratory capacities. Images from (a)to(f) are representative of three independent experiments.

Although inducing a significant reduction of its mRNA non-neoplastic cells RWPE-1 and prostate epithelial cells (PrEC). (Supplementary Figure S3C), shRNA against PDCD4 was slightly Consistently, androgen-independent PC3 and DU145 cells have able to induce vimentin expression in RWPE-1 cells, whereas almost undetectable BTG2 protein, contrary to androgen-depen- fibronectin and E-Cadherin were not affected (Supplementary dent LNCaP and 22RV1, which retain some expression (Figure 3f). Figures S3D and E). However, we cannot exclude that other miR- Thus, the levels of BTG2 in androgen-dependent cancer cells are 21 targets might participate in the observed phenotype. The shift higher than in androgen-independent cell lines, but still lower to a luminal- and EMT-like phenotype was also confirmed in a line than in normal cells, indicating that a progressive loss of BTG2 of normal non-immortalized prostate epithelial cells (Supple- may correlate with tumor progression. We supposed that, in mentary Figures S4A and B) and in tumoral AR-positive LNCaP androgen-dependent prostate cancer cells, basal levels of BTG2 and 22RV1 cells (Supplementary Figures S4C and D). The might represent an anti-oncogenic mechanism to prevent evaluation of miR-21 expression revealed a moderate or strong progression to advanced stages. As LNCaP and 22RV1 cells are upregulation of miR-21 in RWPE-2, PC3 and DU145 cells, whereas both p53-proficient, while PC3 and DU145 are p53-deleted or LNCaP and 22RV1 cells displayed miR-21 levels lower or similar to -mutated, respectively, we reasoned that in androgen-dependent

Oncogene (2013) 1843 – 1853 & 2013 Macmillan Publishers Limited BTG2 loss and miR-21 upregulation V Coppola et al 1847

Figure 3. miR-21 overexpression reproduces the effects of BTG2 knockdown. (a) Luciferase assay on RWPE-1 cells transfected with renilla luciferase and with either empty pGL3 vector or a plasmid carrying the 30-UTR of BTG2 (pGL3 UTR–BTG2), in presence of negative control scramble pre-miR or pre-miR-21. Firefly luciferase activities are normalized over renilla luciferase. Data are reported as mean±s.d. of three independent experiments. (b) Representative fluorescence-activated cell sorting (FACS) analysis of BTG2 expression in RWPE-1 cells transduced with control (TW) or miR-21-overexpressing vector. On the right, histogram depicting the relative BTG2 expression, calculated as average decrease of mean fluorescence intensities (MFI) in three independent experiments. (c) Western blot analysis of CK8-18, AR and p63 expression in control (TW) and miR-21-overexpressing RWPE-1 cells. Tubulin or nucleolin were used as loading controls. (d) IF analysis of vimentin and fibronectin expression in control (TW) and miR-21 overexpressing RWPE-1 cells. The EGFP reporter is shown on the left, whereas the two stainings are merged with DAPI in the right panels. (e) IF analysis of E-Cadherin and quantification of the percentage of E-Cadherin- positive cells in TW and miR-21-overexpressing RWPE-1. In (c), (d) and (e), one representative of three experiments is shown. (f) On the left, real-time PCR analysis of miR-21 expression in different prostate cell lines. RWPE-1 cells were used as reference cell line. On the right, quantification of BTG2 expression by FACS analysis. Values are plotted as MFI of anti-BTG2-incubated cells, normalized over control antibody (Ab)-incubated cells. Values are mean±s.d. of three independent experiments. cancer cells, p53 might be responsible for ensuring a residual shRNA transduction. The partial knockdown of p53 obtained in expression of its target gene BTG2. To test this hypothesis, we these cells was indeed accompanied by a further reduction of interfered with p53 levels in LNCaP and 22RV1 cells by specific BTG2 and by a moderate increase of miR-21 (Supplementary

& 2013 Macmillan Publishers Limited Oncogene (2013) 1843 – 1853 BTG2 loss and miR-21 upregulation V Coppola et al 1848 Figure S5A). However, we cannot exclude that other mechanisms chimera 30-UTR able to bind excess of miR-21 and release might intervene in the complex network governing prostate cancer endogenous BTG2 from posttranscriptional control. The efficacy progression.42 We then analyzed miR-21 in 27 samples of primary of this system was demonstrated by an increase of BTG2 levels in prostate tumor cells and observed that miR-21 was expressed at RWPE-2 cells infected with UTR–BTG2, as compared with empty high or moderate level in about half of the patients (Supplementary vector-transduced cells (Figure 4c). This increase was sufficient to Figure S5B). By setting an arbitrary cutoff level of miR-21 at 2, significantly impact cell growth (Figure 4d), but to a lower extent Gleason grade 47 and pT3N0M0 were observed in 5/11 cases with than the overexpression of BTG2 coding sequence. Consistently miR-21 42, and in 3/16 patients with miR-21 level under 2, with reduced proliferation, RWPE-2 UTR–BTG2 cells showed a suggesting an association of miR-21 upregulation with tumor local marked reduction of Cyclin D1 and pRb (Figure 4e and invasiveness and with advanced forms. The imperfect correlation Supplementary Figure S7C). Importantly, other established miR- between BTG2 loss in all tumors analyzed in our study, and miR-21 21 targets, Sprouty-243 and PTEN,44 were not affected in the overexpression in only half of them may suggest that either RWPE-2 UTR–BTG2 cells (Figure 4f and Supplementary Figure S7D), additional mechanisms (like transcriptional repression or proteaso- suggesting a selective modulation of the targeting of miR-21 on mal degradation) may be involved in BTG2 loss or that negative the 30-UTR of BTG2. It is worthy to note that, the restoration of feedback mechanisms may intervene to reduce miR-21 levels. To BTG2 levels could shift the phenotype of RWPE-2 cells, as shown verify whether a mild increase of miR-21 levels could be functional, by upregulation of p63, downregulation of CK8-18, vimentin and we performed transduction of miR-21 with scalar dilutions of viral fibronectin (Figures 4f and g and Supplementary Figures S7D, E titer (Supplementary Figure S5C). Interestingly, also a low degree of and F). RWPE-2 UTR-BTG2 cells showed an increase of nuclear miR-21 overexpression increased levels of CK8-18 and vimentin, localization of BTG2 and a parallel mild p63 expression with a number of cells fluctuating in a state in which they are (Supplementary Figure S8A). Similar effects were obtained in simultaneously expressed (Supplementary Figure S5C). To clarify PC3 cells transduced with UTR–BTG2 vector (Supplementary this dual role of miR-21, RWPE-1 cells were transfected with miR-21 Figure S8B). To confirm this latter result, we verified that, although precursor oligo, and the expression of CK8-18 and vimentin was at low levels, BTG2 and p63 mRNA could be detected in PC3 cells analyzed after 24 and 96 h. Transfection resulted in massive miR-21 and were also modified by the UTR–BTG2 overexpression overexpression (approximately 500-fold). A marked CK8-18 upre- (Supplementary Figure S8C). The UTR–BTG2 transduction in gulation was observed at 24 h, whereas vimentin overexpression RWPE-1 miR-21 cells was able to reduce EMT traits was detected only at 96 h (Supplementary Figure S5D). This (Supplementary Figure S8D). As the restoration of BTG2 levels suggests that miR-21 shot first induces luminal markers expression, by the UTR–BTG2 overexpression could theoretically be due to whereas a prolonged upregulation enables aberrant expression of depleting other microRNAs capable of binding the 30-UTR of BTG2, both CK8-18 and vimentin. EMT is largely considered a prelude to we verified that similar effects could be reproduced by miR-21 the spread of metastases. Notably, prostate lymphonodal metas- knockdown in RWPE-2 cells by engineering the lentiviral vector tases express CK8-18 (Supplementary Figure S5E), indicating that TW 30-UTR Decoy39 in order to selectively sequester miR-21 (Decoy the induction of both features may be necessary for prostate cancer 21) (Supplementary Figure S8E). progression. In an effort to clarify whether BTG2/miR-21 axis has a role in physiological differentiation, we tried to differentiate in vitro RWPE-1 cells by exposure to dihydrotestosterone for 1 week. We BTG2 knockdown allows orthotopic engraftment of normal observed a partial differentiation of RWPE-1, as documented by a prostate cells decrease of p63 and an increase of CK8-18 (Supplementary Figure We next investigated the consequences of BTG2 loss on prostate S6A). This process was accompanied by a reduction of BTG2 mRNA tumorigenesis. AR overexpression has been shown to enable and by a decrease/delocalization of BTG2 protein (Supplementary tumor formation by previously immortalized normal prostate Figures S6B and C). However, miR-21 levels were not increased but epithelial cells in an orthotopic mouse model.45 Although BTG2 rather decreased in the course of the differentiation process interference did not result in any significant proliferative (Supplementary Figure S6C). This suggests that physiological advantage in vitro (data not shown), we hypothesized that its differentiation is independent of miR-21 increase, and that putative ability to upregulate AR and induce a luminal phenotype may be tumor-suppressor mechanisms might rather reduce miR-21 levels relevant for in vivo engraftment and growth. Therefore, we stably to avoid cell transformation. Thus, miR-21-mediated regulation of knocked down BTG2 in RWPE-1 by inserting shRNAs against BTG2 BTG2 appears relevant only for tumor transformation and cancer in pLox lentiviral vector (shRNA BTG2),46 which includes a cassette progression. for EGFP reporter expression. Levels of BTG2 mRNA were reduced by 50% in RWPE-1 shRNA BTG2 cells (data not shown). Control and BTG2-interfered RWPE-1 cells were then inoculated in the left Derepression of BTG2 mRNA can reduce luminal phenotype anterior prostate of 10 immunodeficient mice for each group, and and EMT dihydrotestosterone pellets were subcutaneously implanted to We then aimed at investigating the consequences of BTG2 ensure optimal hormonal support. Mice were killed 7 weeks after restoration in RWPE-2 cells. Consistently with miR-21 increase, we surgery, and prostates observed at the stereomicroscope to observed an accumulation of the luminal markers CK8-18 and AR evaluate the engraftment of human prostate cells on the basis of in these early-tumoral cells (Supplementary Figure S7A). The EGFP signal. The absence or presence of an EGFP signal was then transduction of these cells with a lentiviral construct carrying only used to calculate the rate of engraftment. Significantly, prostates BTG2 coding sequence (TW BTG2) resulted in dramatic growth inoculated with control RWPE-1 pLox cells did not show any arrest (Figure 4a). Similar effects were observed upon BTG2 fluorescent signal in 7/10 cases, whereas in 3/10 prostates, a weak overexpression in PC3 cells (Supplementary Figure S7B). In an EGFP positivity was detected. Conversely, only 2/10 prostates effort to confirm the role of BTG2 in basal-luminal shift and EMT in inoculated with RWPE-1 shRNA BTG2 were negative for EGFP, a tumoral model, we devised a system to obtain a selective and whereas in 8/10 cases, EGFP signal revealed a striking capacity of more physiological BTG2 restoration by interfering with miR- these cells to grow in an orthotopic site, as demonstrated by a mediated control of BTG2 levels. We produced a lentiviral vector, strong EGFP signal in 6/10 prostate and a weaker positivity in 2/10 called TW–30-UTR–BTG2 (UTR–BTG2), in which a region of the (Figure 5a). Consistently, hematoxylin and eosin staining of 30-UTR of BTG2, including miR-21 binding sites, is inserted in the xenografts sections revealed a massive presence of RWPE-1 30-UTR of the enhanced green fluorescent protein (EGFP). shRNA BTG2 cells invading the murine prostate (Figure 5b). Thus, (Figure 4b). This vector produces a stable EGFP transgene with a BTG2 loss may be per se an oncogenic mechanism.

Oncogene (2013) 1843 – 1853 & 2013 Macmillan Publishers Limited BTG2 loss and miR-21 upregulation V Coppola et al 1849

Figure 4. Derepression of BTG2 mRNA can reduce luminal phenotype and EMT. (a) Histogram depicting cell number fold increase of control (TW) and BTG2 overexpressing (TW BTG2) RWPE-2 cells between plating day (day 1) and day 4. Values were obtained by trypan blue counting and evaluation of cells’ metabolic capacity by Cell Titer Blue assay and represent mean±s.d. of three independent experiments. ***Po0.001. (b) Scheme of TW 30-UTR–BTG2 (briefly, UTR–BTG2) lentiviral vector for BTG2 mRNA derepression. LTR, long terminal repeats; CMV, cytomegalovirus promoter; EGFP, enhanced green fluorescent protein; hPGK, human phosphoglycerate kinase promoter; PURO R, puromycin resistance gene. (c) Representative fluorescence-activated cell sorting analysis of BTG2 expression in control (UTR) and BTG2-derepressed (UTR–BTG2) RWPE-2 cells. On the right, histogram plotting the average MFI±s.d. of three independent experiments. Values were obtained as described for Figure 3e. *Po0.05. (d) Histogram depicting cell number fold increase of control (UTR) and BTG2-derepressed (UTR–BTG2) RWPE- 2 cells from plating day (day1) till day 4. Values were obtained by trypan blue counting and Cell Titer Blue assay and represent mean±s.d. of three independent experiments. **Po0.01. (e) Western blot analysis of cyclin D1 and pRb in RWPE-1 and in RWPE-2 cells transduced with UTR or UTR–BTG2 vectors. (f) Western blot analysis of p63, Sprouty-2 and PTEN in UTR and BTG2-derepressed (UTR–BTG2) RWPE-2 cells. In (e) and (f), actin was used as a loading control. (g) Cytofluorimetric analysis of CK8-18, vimentin and fibronectin in UTR and UTR–BTG2-transduced RWPE-2 cells. In (e), (f) and (g), one representative of three experiments is shown.

& 2013 Macmillan Publishers Limited Oncogene (2013) 1843 – 1853 BTG2 loss and miR-21 upregulation V Coppola et al 1850

Oncogene (2013) 1843 – 1853 & 2013 Macmillan Publishers Limited BTG2 loss and miR-21 upregulation V Coppola et al 1851 DISCUSSION to brake proliferation in response to cellular growth signals. In this study, we demonstrated that BTG2 targeting by miR-21 Consistently, functional studies indicated that ectopic expression promotes the shift of prostatic basal cells towards the luminal of BTG2 caused a significant inhibition of the AR-mediated phenotype coupled with the appearance of EMT features and transcriptional activity, an attenuation of prostate-specific antigen 50 increased tumorigenic activity. It is generally accepted that levels and a decreased growth of prostate cancer cells. This is prostate cancer cells derive from the basal compartment, but also consistent with our observation showing luminal expression express luminal markers.47 Given that the tumor suppressor BTG2 of BTG2 in acini dispersed in tumor areas, having been previously is differently expressed within the epithelial populations of reported overexpression of BTG2 in the hyperproliferative lesions prostate cells (basal versus luminal) and lost in prostate cancer, simple atrophy, post-atrophic hyperplasia and proliferative 20 we investigated whether BTG2 downregulation could be inflammatory atrophy. Interestingly, the luminal phenotype implicated in cell transformation and cancer progression. To this induced by BTG2 knockdown and miR-21 overexpression was aim, we employed a strategy of BTG2 interference in RWPE-1, a accompanied by the acquisition of mesenchymal-like traits. EMT is normal non-tumorigenic cell line composed by two a trans-differentiation process that illustrates the plasticity of subpopulations that differ in size and differentiation status.48 We cancer cells and is thought to mark a key step towards the spread 12 showed that BTG2 downregulation can shift the balance between of metastases. Interestingly, several reports emphasize that these subpopulations, leading to the accumulation of large AR- circulating tumor cells coexpress epithelial and mesenchymal 51 positive cells expressing high CK8-18 and low p63 levels. In a markers. In our model, this phenotype may be explained by the widely accepted model of prostate development and neoplastic influence of BTG2 on the levels of the tumor suppressor DNp63. 35 transformation, a small stem cell population located in the basal However, given the inhibitory effect of BTG2 on RAS activity, we cell layer gives rise to all epithelial cell lineages encountered in the cannot exclude that an enhanced ERK signaling due to BTG2 loss normal, hyperplastic and neoplastic tissues.3 The differentiating may be involved in inducing this phenotype. Interestingly, the process can be induced by circulating androgens, largely effects observed upon BTG2 knockdown in non-tumorigenic depending on the presence of androgen-responsive target RWPE-1 cells could be reverted in early-tumoral RWPE-2 cells cells.49 Accordingly, the abnormal growth of the secretory upon derepression of BTG2 mRNA. A vector to overexpress the 30- epithelium in benign prostate hyperplasia may be related to an UTR of a given gene to compete with endogenous mRNA for increase in the total number of androgen-responsive cells. Berger microRNA binding is an original tool to finely modulate gene et al.45 demonstrated that the overexpression of AR was sufficient expression. This can be especially useful in the study of tumor to convert immortalized prostate epithelial cells into differentiated suppressor genes like BTG2. In these cases, a careful dosage of secretory cells that recapitulated the phenotype associated with gene expression may be necessary to unveil novel functions human prostate cancer and were able to grow in an orthotopic normally masked by the dramatic effects on cell growth and murine model. Similar results were obtained in our model of BTG2 viability due to the artificial overexpression of the coding knockdown in RWPE-1 cells, whose growth occurred upon sequence. orthotopic, but not subcutaneous injection (data not shown). Therefore, dysregulation of the miR-21/BTG2 axis may not be sufficient to make normal cells tumorigenic without the MATERIALS AND METHODS contribution of a suitable microenvironment. We demonstrated Flow cytometry and immunofluorescence that in prostate cells, the overexpression of the established oncogene miR-21 can suppress BTG2 expression and reproduce For intracellular flow cytometry staining, cells were fixed with 2% paraformaldehyde for 20 min and permeabilized with phosphate-buffered the effects observed upon BTG2 interference. Interestingly, the saline 0.1% Triton X-100 for 3 min at room temperature. Cells were then overexpression of miR-21 in AR-positive LNCaP and 22RV1 cells incubated for 30 min on ice with primary antibody diluted in phosphate- further increases the amount of receptor, possibly increasing buffered saline 1% bovine serum albumin. The antibodies used include: sensitivity to lower levels of androgens. This finding is in line with anti BTG2 (1:40; sc-30342; Santa Cruz Biotechnology, Santa Cruz, CA, USA), previous observations showing that the inhibition of miR-21 can anti p63 (1:200, 559951; BD Pharmingen, San Jose, CA, USA), anti reduce androgen-induced prostate cancer cell proliferation, and cytokeratin 8-18 (1:50; sc-52325; Santa Cruz), anti vimentin (1:100; 3295; that elevated expression of miR-21 in prostate cancer cells Cell Signalling, Danvers, MA, USA) and anti fibronectin (1:100; F7387; enhances tumor growth in vivo and allows androgen-dependent Sigma-Aldrich, St Louis, MO, USA). Cells were washed twice and then tumors to overcome castration-mediated growth arrest.30 incubated for 30 min on ice with secondary antibodies. For BTG2 detection, PE-conjugated donkey anti-goat antibody (1:200; 705-001-003; Jackson Furthermore, serum miR-21 levels have been reported to ImmunoResearch, West Grove, PA, USA) was used. For the other antibodies correlate with those of the prostate-specific antigen in patients Alexa Fluor 647-conjugated goat anti-mouse and goat anti-rabbit 27 with androgen-dependent and -independent prostate cancer. In antibodies were used (Invitrogen, Carlsbad, CA, USA). For E-Cadherin addition, a positive feedback loop can be designed by joining our staining, cells were trypsinized, incubated in a medium overnight on a observations with the reported transactivation of miR-21 promoter rocker platform to enable regeneration of the protein, and then stained by AR.30 Therefore, we envisage that miR-21 induction may sustain with PE-conjugated anti E-Cadherin monoclonal antibody according to the AR signaling, participating in molecular mechanisms involved in manufacturer’s instructions (FAB18381P; R&D Systems, Minneapolis, MN, resisting endocrine therapy. In light of the above, low or moderate USA). Stainings were analyzed with BD FACSCanto flow cytometer (Becton, levels of miR-21 detected in some cell lines, as well as in a Dickinson and Company, San Jose, CA, USA). For immunofluorescence analysis, cells (3–6 Â 104 on each 15-mm polylysine-coated coverslip subgroup of primary prostate tumor cells, may represent a placed in the 24-well plate) were seeded in culture medium. Fixation and strategy to prevent progression to castration resistance. It is permeabilization were performed as described for the cytofluorimetric conceivable that low levels of miR-21 and an induction of BTG2,a analysis. E-Cadherin immunofluorescence was performed with an antibody p53 responsive gene, might represent a feedback mechanism able recognizing the cytoplasmic domain of the molecule (1:200; 610181; BD

Figure 5. BTG2 knockdown allows orthotopic engraftment of normal prostate cells. (a) Transmitted-light and fluorescent stereomicroscopic pictures of mice anterior prostates inoculated with control (pLox) or BTG2-interfered (shRNA BTG2) RWPE-1 cells. Prostates (3/10) for each group are shown. Magnification Â 12.5. Below, pie chart summarizing the % of engraftment of pLox and shRNA BTG2 RWPE-1 cells in the murine orthotopic model. EGFP signal was classified as negative, weak or strong. The statistical significance of the frequency distribution was evaluated by w2-test (Po0.0001) (b) Hematoxylin and eosin staining of cryostatic slides of mouse prostates inoculated with pLox and shRNA BTG2 RWPE-1 cells. One pLox and two representative shRNA BTG2 samples are shown.

& 2013 Macmillan Publishers Limited Oncogene (2013) 1843 – 1853 BTG2 loss and miR-21 upregulation V Coppola et al 1852 Pharmingen). Primary antibodies were diluted at the previously indicated CONFLICT OF INTEREST concentrations in phosphate-buffered saline 3% bovine serum albumin, The authors declare no conflict of interest. 0.05% Triton X-100 and incubated overnight at 4 1C. After extensive washing, coverslips were incubated for 45 min at room temperature with the appropriate secondary antibodies diluted 1:2000. For vimentin– fibronectin or vimentin–CK8-18 co-staining, Alexa Fluor 555- and 647- ACKNOWLEDGEMENTS conjugated antibodies were used. For BTG2-p63 co-staining, 647- We thank F. Tirone and L. Leonardi for providing non-commercial His-tagged anti- conjugated donkey anti goat and TRITC-conjugated rabbit anti mouse BTG2 antibody, S. Soddu for kindly providing pSUPER-p53plasmid for p53 antibodies were used. After extensive washing, the nuclei were stained interference, T. Merlino for manuscript editing, and G. Loreto for graphical assistance. with DAPI (40,60-diamidino-2-phenylindole) for 20 min. Cytoskeletal actin This work was supported by the Italian Health Ministry with ‘Under forty researchers was stained with Alexa Fluor 488-conjugated Phalloidin (200U/staining; 2007’ and Italy-USA microRNA program to DB, and by the Italian Association for A12379; Invitrogen). Coverslips were mounted on microscope slides and Cancer Research (AIRC) and ‘Fondazione Roma’ support to RDM. sealed. Primary antibody-free controls were produced. Images were collected on a fluorescence confocal microscope (Olympus FV1000; Olympus, Tokyo, Japan). For the analysis of prostate specimens, samples REFERENCES obtained from radical prostatectomy were snap-frozen in optimal cutting 1 Gronberg H. Prostate cancer epidemiology. Lancet 2003; 361: 859–864. temperature compound (OCT) and stored at 80 1C. Cryostatic sections À 2 Wang Y, Hayward S, Cao M, Thayer K, Cunha G. Cell differentiation lineage in the were obtained with a cryomicrotome (Kriostat 1720 MGW; Leitz, Melville, prostate. Differentiation 2001; 68: 270–279. NY, USA). 3 Bonkhoff H, Stein U, Remberger K. The proliferative function of basal cells in the normal and hyperplastic human prostate. Prostate 1994; 24: 114–118. Transfection 4 Nagle RB, Ahmann FR, McDaniel KM, Paquin ML, Clark VA, Celniker A. Cytokeratin characterization of human prostatic carcinoma and its derived cell lines. Cancer RWPE-1 cells were transfected with 30 nM of either negative control or pre- miR-21 (Ambion/Applied Biosystems, Foster City, California, CA, USA), Res 1987; 47: 281–286. Hs_BTG2_6, Hs_KRAS_3 plus Hs_KRAS_8 FlexiTube siRNAs (Qiagen, Hilden, 5 Sherwood ER, Berg LA, Mitchell NJ, McNeal JE, Kozlowski JM, Lee C. Differential Germany) and p63 siRNAs, using the Hiperfect transfection reagent cytokeratin expression in normal, hyperplastic and malignant epithelial cells from (Qiagen). RWPE-2 or RWPE-1 miR-21 cells were transfected with expression human prostate. J Urol 1990; 143: 167–171. constructs pCS2, pCS2 DNp63 and pCS2 TAp63 using Lipofectamine 2000 6 Lam JS, Reiter RE. Stem cells in prostate and prostate cancer development. Urol (Invitrogen), according to the manufacturer’s instructions. p63 siRNAs and Oncol 2006; 24: 131–140. expression constructs are referenced in Adorno et al.11 Unless differently 7 McDonnell TJ, Troncoso P, Brisbay SM, Logothetis C, Chung LW, Hsieh JT et al. indicated, cells were collected 72 h after transfection for further analysis. Expression of the protooncogene bcl-2 in the prostate and its association with emergence of androgen-independent prostate cancer. Cancer Res 1992; 52: 6940–6944. Generation of lentiviral vectors and gene transfer 8 Signoretti S, Waltregny D, Dilks J, Isaac B, Lin D, Garraway L et al. p63 is a prostate For TWmiR-21 generation, miR-21 precursor DNA was PCR amplified from basal cell marker and is required for prostate development. Am J Pathol 2000; human genomic DNA with the following primers: forward 50-cacactctgtcg 157: 1769–1775. tatctg-30; reverse 50-gactatccccatttctcc-30. The 618 bp fragment was 9 Signoretti S, Pires MM, Lindauer M, Horner JW, Grisanzio C, Dhar S et al. p63 subcloned into the lentiviral vector TW under the control of the regulates commitment to the prostate cell lineage. Proc Natl Acad Sci USA 2005; cytomegalovirus promoter. For BTG2 overexpression, the 476 bp of BTG2 102: 11355–11360. coding sequence were cloned in the TW vector under the control of the 10 Barbieri CE, Tang LJ, Brown KA, Pietenpol JA. Loss of p63 leads to increased cell cytomegalovirus promoter. Primers to amplify the coding sequence were: migration and up-regulation of genes involved in invasion and metastasis. Cancer forward 50-acatgagccacgggaag-30; reverse 50-gaagggcctagctggaga-30. For Res 2006; 66: 7589–7597. TW–30-UTR–BTG2 generation, an 808 bp fragment of BTG2 30-UTR including 11 Adorno M, Cordenonsi M, Montagner M, Dupont S, Wong C, Hann B et al. A miR-21 binding site was PCR amplified from human genomic DNA with the mutant-p53/Smad complex opposes p63 to empower TGFbeta-induced metas- following primers: forward 50-ggaaaggttgctgtctgtctc-30; reverse 50-ggtgta tasis. Cell 2009; 137: 87–98. catttgtccataagctg-30.TW30-UTR Decoy 21 was generated by synthesizing 12 Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transition. J Clin two antisense sequences of miR-21 separated by a stem-loop spacer. The Invest 2009; 119: 1420–1428. fragments were subcloned in the EGFP 30-UTR of the TW30 vector described 13 Melamed J, Kernizan S, Walden PD. Expression of B-cell translocation gene 2 in Bonci et al.,39 endowed with a puromycin-resistance cassette. For the protein in normal human tissues. Tissue Cell 2002; 34: 28–32. pLox shRNA BTG2 and PDCD4 generation, the following sequences were 14 Rouault JP, Falette N, Guehenneux F, Guillot C, Rimokh R, Wang Q et al. Identi- chosen for shRNA design according to the instructions reported in fication of BTG2, an antiproliferative p53-dependent component of the DNA Rubinson et al.46: for BTG2, 50-tcaaccacaagatggaccc-30;50-ggagaaaagtg damage cellular response pathway. Nat Genet 1996; 14: 482–486. gaaagcta-30;50-gaaacaaagctgtgaatca-30; for PDCD4 50-atcctgataacttaagt-30; 15 Guardavaccaro D, Corrente G, Covone F, Micheli L, D’Agnano I, Starace G et al. 50-aactgtgtttatgaaac-30. About 50 nucleotide long stem-loop sense and Arrest of G(1)-S progression by the p53-inducible gene PC3 is Rb dependent antisense oligonucleotides were synthesized and 60 pmol of each oligo and relies on the inhibition of cyclin D1 transcription. Mol Cell Biol 2000; 20: were in vitro annealed by incubating the reaction in the annealing buffer 1797–1815. (100 mM potassium acetate, 30 mM HEPES–KOH pH 7.4, 2 mM magnesium 16 Lim IK. TIS21 (/BTG2/PC3) as a link between ageing and cancer: cell cycle reg- acetate) for 4 min at 95 1C, 10 min at 70 1C and by slowly decreasing the ulator and endogenous cell death molecule. J Cancer Res Clin Oncol 2006; 132: temperature down to 4 1C(11C/min). Lentiviral particles were produced as 417–426. previously described.52 Cells were transduced with 1 Â 106 viral 17 Boiko AD, Porteous S, Razorenova OV, Krivokrysenko VI, Williams BR, Gudkov AV. transducing units per ml, unless differently indicated. For BTG2 and A systematic search for downstream mediators of tumor suppressor function of PDCD4 stable interference, viruses encoding different shRNA sequences p53 reveals a major role of BTG2 in suppression of Ras-induced transformation. were mixed in equal volumes. EGFP was used as a reporter and EGFP Genes Dev 2006; 20: 236–252. bright cells were sorted with BD FACSAria (Becton, Dickinson and 18 Struckmann K, Schraml P, Simon R, Elmenhorst K, Mirlacher M, Kononen J et al. Company). The selection of cells transduced with TW 30-UTR and TW 30- Impaired expression of the cell cycle regulator BTG2 is common in clear cell renal UTR–BTG2 was performed by applying 1 mg/ml puromycin for 48 h. For p53 cell carcinoma. Cancer Res 2004; 64: 1632–1638. interference, constructs described in Cecchinelli et al.53 were used. 19 Kawakubo H, Brachtel E, Hayashida T, Yeo G, Kish J, Muzikansky A et al. Loss of B-cell translocation gene-2 in estrogen receptor-positive breast carcinoma is associated with tumor grade and overexpression of cyclin d1 protein. Cancer Res Statistical analysis 2006; 66: 7075–7082. Plotted data are presented as the mean þ s.d. of at least three experiments. 20 Ficazzola MA, Fraiman M, Gitlin J, Woo K, Melamed J, Rubin MA et al. Anti- Statistical significance was calculated by the Student’s t-test analysis. proliferative B cell translocation gene 2 protein is down-regulated post-tran- Statistical significance of in vivo results was calculated by the w2-test. The scriptionally as an early event in prostate carcinogenesis. Carcinogenesis 2001; 22: immunofluorescence, western blot and fluorescence-activated cell sorting 1271–1279. results represent the three experiments. 21 Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 2004; For additional information see Supplementary Materials and Methods. 116: 281–297.

Oncogene (2013) 1843 – 1853 & 2013 Macmillan Publishers Limited BTG2 loss and miR-21 upregulation V Coppola et al 1853 22 Bartel DP, Chen CZ. Micromanagers of gene expression: the potentially wide- 39 Bonci D, Coppola V, Musumeci M, Addario A, Giuffrida R, Memeo L et al. The miR- spread influence of metazoan microRNAs. Nat Rev Genet 2004; 5: 396–400. 15a-miR-16-1 cluster controls prostate cancer by targeting multiple oncogenic 23 Calin GA, Croce CM. MicroRNA signatures in human cancers. Nat Rev Cancer 2006; activities. Nat Med 2008; 14: 1271–1277. 6: 857–866. 40 Frankel LB, Christoffersen NR, Jacobsen A, Lindow M, Krogh A, Lund AH. Pro- 24 Coppola V, De Maria R, Bonci D. MicroRNAs and prostate cancer. Endocr Relat grammed cell death 4 (PDCD4) is an important functional target of the microRNA Cancer 2010; 17: F1–17. miR-21 in breast cancer cells. J Biol Chem 2008; 283: 1026–1033. 25 Esquela-Kerscher A, Slack FJ. Oncomirs - microRNAs with a role in cancer. Nat Rev 41 Wang Q, Sun Z, Yang HS. Downregulation of tumor suppressor Pdcd4 promotes Cancer 2006; 6: 259–269. invasion and activates both beta-catenin/Tcf and AP-1-dependent transcription in 26 Volinia S, Calin GA, Liu CG, Ambs S, Cimmino A, Petrocca F et al. A microRNA colon carcinoma cells. Oncogene 2008; 27: 1527–1535. expression signature of human solid tumors defines cancer gene targets. Proc 42 Jalava SE, Urbanucci A, Latonen L, Waltering KK, Sahu B, Ja¨nne OA et al. Andro- Natl Acad Sci USA 2006; 103: 2257–2261. gen-regulated miR-32 targets BTG2 and is overexpressed in castration-resistant 27 Zhang HL, Yang LF, Zhu Y, Yao XD, Zhang SL, Dai B et al. Serum miRNA-21: prostate cancer. Oncogene 2012; 23. doi:10.1038/onc.2011.624 (Epub ahead elevated levels in patients with metastatic hormone-refractory prostate cancer of print). and potential predictive factor for the efficacy of docetaxel-based chemotherapy. 43 Sayed D, Rane S, Lypowy J, He M, Chen IY, Vashistha H et al. MicroRNA-21 targets Prostate 2011; 71: 326–331. Sprouty2 and promotes cellular outgrowths. Mol Biol Cell 2008; 19: 3272–3282. 28 Hao Y, Zhao Y, Zhao X, He C, Pang X, Wu TC et al. Improvement of prostate cancer 44 Meng F, Henson R, Wehbe-Janek H, Ghoshal K, Jacob ST, Patel T. MicroRNA-21 detection by integrating the PSA test with miRNA expression profiling. Cancer regulates expression of the PTEN tumor suppressor gene in human hepatocellular Invest 2011; 29: 318–324. cancer. Gastroenterology 2007; 133: 647–658. 29 Li T, Li D, Sha J, Sun P, Huang Y. MicroRNA-21 directly targets MARCKS and 45 Berger R, Febbo PG, Majumder PK, Zhao JJ, Mukherjee S, Signoretti S et al. promotes apoptosis resistance and invasion in prostate cancer cells. Biochem Androgen-induced differentiation and tumorigenicity of human prostate epi- Biophys Res Commun 2009; 383: 280–285. thelial cells. Cancer Res 2004; 64: 8867–8875. 30 Ribas J, Ni X, Haffner M, Wentzel EA, Salmasi AH, Chowdhury WH et al. miR-21: an 46 Rubinson DA, Dillon CP, Kwiatkowski AV, Sievers C, Yang L, Kopinja J et al. A androgen receptor-regulated microRNA that promotes hormone-dependent and lentivirus-based system to functionally silence genes in primary mammalian cells, hormone-independent prostate cancer growth. Cancer Res 2009; 69: 7165–7169. stem cells and transgenic mice by RNA interference. Nat Genet 2003; 33: 401–406. 31 Passeri D, Marcucci A, Rizzo G, Billi M, Panigada M, Leonardi L et al. Btg2 enhances 47 Maitland NJ, Frame FM, Polson ES, Lewis JL, Collins AT. Prostate cancer stem cells: retinoic acid-induced differentiation by modulating histone H4 methylation and do they have a basal or luminal phenotype? Hormones Cancer 2011; 2: 47–61. acetylation. Mol Cell Biol 2006; 26: 5023–5032. 48 Tokar EJ, Ancrile BB, Cunha GR, Webber MM. Stem/progenitor and intermediate 32 Kawamura-Tsuzuku J, Suzuki T, Yoshida Y, Yamamoto T. Nuclear localization of cell types and the origin of human prostate cancer. Differentiation 2005; 73: Tob is important for regulation of its antiproliferative activity. Oncogene 2004; 23: 463–473. 6630–6638. 49 Bonkhoff H, Remberger K. Differentiation pathways and histogenetic aspects of 33 Bello D, Webber MM, Kleinman HK, Wartinger DD, Rhim JS. Androgen responsive normal and abnormal prostatic growth: a stem cell model. Prostate 1996; 28: adult human prostatic epithelial cell lines immortalized by human papillomavirus 98–106. 18. Carcinogenesis 1997; 18: 1215–1223. 50 Hu XD, Meng QH, Xu JY, Jiao Y, Ge CM, Jacob A et al. BTG2 is an LXXLL-dependent 34 Webber MM, Bello D, Quader S. Immortalized and tumorigenic adult human co-repressor for androgen receptor transcriptional activity. Biochem Biophys Res prostatic epithelial cell lines: characteristics and applications Part 2. Tumorigenic Commun 2011; 404: 903–909. cell lines. Prostate 1997; 30: 58–64. 51 Armstrong AJ, Marengo MS, Oltean S, Kemeny G, Bitting RL, Turnbull JD et al. 35 Buganim Y, Solomon H, Rais Y, Kistner D, Nachmany I, Brait M et al. p53 Regulates Circulating tumor cells from patients with advanced prostate and breast cancer the Ras circuit to inhibit the expression of a cancer-related gene signature by display both epithelial and mesenchymal markers. Mol Cancer Res 2011; 9: various molecular pathways. Cancer Res 2010; 70: 2274–2284. 997–1007. 36 Hatley ME, Patrick DM, Garcia MR, Richardson JA, Bassel-Duby R, van Rooij E et al. 52 Bonci D, Cittadini A, Latronico MV, Borello U, Aycock JK, Drusco A et al. ‘Advanced’ Modulation of K-Ras-dependent lung tumorigenesis by microRNA-21. Cancer cell generation lentiviruses as efficient vectors for cardiomyocyte gene transduction 2010; 18: 282–293. in vitro and in vivo. Gene Ther 2003; 10: 630–636. 37 Medina PP, Nolde M, Slack FJ. OncomiR addiction in an in vivo model of micro- 53 Cecchinelli B, Lavra L, Rinaldo C, Iacovelli S, Gurtner A, Gasbarri A et al. Repression RNA-21-induced pre-B-cell lymphoma. Nature 2010; 467: 86–90. of the antiapoptotic molecule galectin-3 by homeodomain-interacting protein 38 Liu M, Wu H, Liu T, Li Y, Wang F, Wan H et al. Regulation of the cell cycle gene, kinase 2-activated p53 is required for p53-induced apoptosis. Mol Cell Biol 2006; BTG2, by miR-21 in human laryngeal carcinoma. Cell Res 2009; 19: 828–837. 26: 4746–4757.

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

& 2013 Macmillan Publishers Limited Oncogene (2013) 1843 – 1853