Oncogene (2013) 32, 2315–2324 & 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13 www.nature.com/onc

ORIGINAL ARTICLE Aberrant expression of the neuronal-specific DCDC2 promotes malignant phenotypes and is associated with prostate cancer progression

N Longoni1, P Kunderfranco1, S Pellini1, D Albino1, M Mello-Grand2, S Pinton1, G D’Ambrosio3, M Sarti1, F Sessa3,4, G Chiorino2, CV Catapano1 and GM Carbone1

By integrating profiling and immunohistochemical data with functional experiments in cell lines in this study we show for the first time that (DCX) domain containing 2 (DCDC2), a protein belonging to the DCX family and involved in neuronal cell migration, is aberrantly expressed in prostate tumors whereas absent in normal prostate. Furthermore, in patients treated with radical prostatectomy, high levels of DCDC2 RNA were significantly associated with increased biochemical relapse (LogRank Mantel-Cox ¼ 0.012). Mechanistically, we found that the ETS transcription factor ESE3/EHF, which is expressed in normal prostate and frequently lost in prostate tumors, maintained DCDC2 repressed by binding to a novel identified ETS binding site in the gene promoter. Consistently, in prostate tumors and in cellular models of gain and loss of ESE3/EHF, the expression of DCDC2 and ESE3/ EHF were inversely correlated. In prostate cancer cells, DCDC2 colocalized with and promoted cell migration and resistance to the -targeting drug taxol. Collectively, this study establishes DCDC2 as a novel ESE3/EHF oncogenic target in prostate cancer. These findings may be relevant for the clinical management of prostate cancer as DCDC2 may signal tumors more prone to relapse and resistant to taxol treatment.

Oncogene (2013) 32, 2315–2324; doi:10.1038/onc.2012.245; published online 25 June 2012 Keywords: prostate cancer; DCDC2; doublecortin domain containing 2 protein; ESE3/EHF; ETS; cell migration; taxol resistance

INTRODUCTION is composed of eleven prologs in humans and in mice. DCX Prostate cancer is the most common cancer and a leading cause are involved in the neuronal migratory activity during 13–15 of cancer death in western countries.1 Mortality is mostly due to cortex development. Mutations in the human DCX gene, the metastatic disease and development of castration-resistant first characterized gene of the family, result in abnormal neuronal prostate cancer. Improving the management of prostate cancer migration, epilepsy, mental retardation and cause double cortex 11 will require early diagnosis and appropriate therapeutic decisions syndrome and in humans. Downregulation of based on a better understanding of the underlying molecular DCX by RNA interference in animal models leads to neuronal events affecting the progression of the disease. The androgen migration disorders similar to those seen in the brains of dyslexic 13,16 receptor has an important role in prostate cancer initiation and individuals. DCDC2 is a brain-specific protein interacting with 13,15–18 progression to castration resistance.1 In addition to androgen microtubules and involved in neuronal migratory activity receptor, transcription factors of the ETS gene family have Mutations in the DCDC2 gene have been associated with dyslectic 13 emerged as important elements in the pathogenesis of prostate disabilities due to abnormal neuronal migration. Our cancer.2–4 Chromosomal translocations and gene rearrangements finding of high expression of DCDC2 in prostate tumors was leading to ectopic expression of ETS occur in about intriguing because the level of DCDC2 is negligible in many 50% of primary tumors.2–4 Translocated ETS factors have normal tissues including prostate. The restricted pattern of oncogenic activity promoting cell transformation, proliferation expression of DCDC2 and its specialized function in neuronal and survival.5–9 We reported recently that the epithelial-specific brain migration suggested that DCDC2, when ectopically ETS factor EHF/ESE3, which is expressed in normal prostate expressed in prostate epithelial cells, could confer abnormal epithelial cells, has tumor-suppressor function and is frequently properties contributing to cell transformation and tumorigenesis. downregulated in prostate tumors due at least in part to Consistent with this hypothesis, we found that high expression of epigenetic events.10 DCDC2 was associated with increased biochemical relapse in In this study, we identified for the first time that the gene patients after radical prostatectomy. Mechanistically, DCDC2 encoding the neuronal-specific doublecortin (DCX) domain expression was negatively regulated by ESE3/EHF and containing 2 (DCDC2) protein as an ETS target gene that is reactivated upon loss of ESE3/EHF in prostate cancer cells. upregulated as consequence of ESE3/EHF loss in prostate tumors. Functionally, ectopic expression of DCDC2 affected migration DCDC2 belongs to the DCX family of proteins characterized by the and resistance to taxol in prostate cancer cells. Collectively, this presence of DCX domains.11,12 The DCX repeat gene family study identifies a novel oncogenic target regulated by the ETS

1Institute of Oncology Research (IOR) and Oncology Institute of Southern Switzerland (IOSI), Bellinzona, Switzerland; 2Laboratory of Cancer Genomics, Fondazione Edo ed Elvo Tempia Valenta, Biella, Italy; 3IRCCS Multimedica, Milan, Italy and 4Department of Pathology, University of Insubria, Varese, Italy. Correspondence: Dr GM Carbone, Institute of Oncology Research (IOR), Oncology Institute of Southern Switzerland (IOSI), Via Vela 6, Bellinzona 6500, Switzerland. E-mail: [email protected] Received 26 January 2012; revised 30 March 2012; accepted 23 April 2012; published online 25 June 2012 DCDC2 is overexpressed in prostate cancer N Longoni et al 2316 factor ESE3/EHF in prostate tumors with potential relevance as inflammation (Figure 1e). The list of positively correlated genes diagnostic, prognostic and predictive biomarker. included known oncogenic targets with tumor promoting functions such as MMP19, IL1B and chemokine receptor CXCR4 among others. Intriguingly, the genes negatively correlated to DCDC2 were associated prevalently to chromatin and nucleosome RESULTS assembly. Although this might suggest an association with altered DCDC2 is aberrantly expressed in prostate cancers epigenetic control mechanisms, understanding the significance of To identify genes that mark the molecular transition from benign this finding would require further studies. prostate to invasive carcinoma, we performed differential profile analysis on a previously published gene expression profiling data set of normal prostate and primary Expression of DCDC2 is negatively regulated by ESE3/EHF prostate cancers.19 We focused particularly on ectopically To validate the finding of aberrant expression of DCDC2 in expressed genes, that is, genes that are not expressed in adult prostate tumors at the protein level, we performed immunohis- normal prostatic tissue and whose expression would indicate the tochemical staining with an anti-DCDC2 antibody on tissue activation of an abnormal differentiation program in tumor cells. microarrays containing specimens of normal prostate (n ¼ 47) In addition to their biological relevance in the process of and matching organ-confined prostate tumors (n ¼ 50). DCDC2 malignant transformation, such genes could be relevant as immunostaining was negative in 90% of the cases and weakly potential diagnostic and prognostic biomarkers and targets for present only in 10% of normal prostate samples, confirming that therapeutic intervention. Among the significantly differentially the protein is generally not expressed in normal adult tissues expressed genes, we found DCDC2 (FDR ¼ 0.002), a gene (Figures 2a,b). About 30% of prostate tumors were negative, while encoding a neuronal-specific protein involved in neuronal 28 and 42% of the cases were, respectively, weakly and strongly migration and brain development and not expressed in most positive for DCDC2 (Figures 2a,b). DCDC2 immunostaining was adult tissues.17 DCDC2 was significantly increased in prostate prevalently cytoplasmic and more abundant in the perinuclear tumors compared with normal prostate samples (Figure 1a, left region both in normal and tumor samples (Figure 2b). Thus, panel). In the microarray data set about half (55%) of tumors had immunohistochemical assessment of DCDC2 level reinforced the a X2 increase of DCDC2 mRNA compared with normal prostate RNA data confirming that DCDC2 is absent in normal prostate and and 17% of tumors had an increase of X3 fold (Figure1a, right is expressed in a high percentage of prostate tumors. panel). To confirm this finding, we measured DCDC2 mRNA by The absence of DCDC2 in normal prostate and its frequent re- quantitative real-time-PCR (qRT–PCR) in an independent set of expression in prostate tumors suggested that the event could be formalin-fixed paraffin embedded (FFPE) primary prostate cancer linked to malignant transformation and be part of the dediffer- samples. DCDC2 was expressed in about 60% of the tumors, entiation program induced by oncogenic events in prostate although it was undetectable by qRT–PCR in the remaining cases epithelial cells. Therefore, we were interested in understanding (Figure 1b), consistent with the percentage of DCDC2 positive the mechanisms controlling DCDC2 expression in these cells. tumors in the microarray data set. Interestingly, correlation analysis in normal and prostate cancer To assess the clinical relevance of this finding, we determined microarray data revealed that the ETS transcription factor ESE3/ whether expression of DCDC2 was associated with clinical EHF was inversely correlated to DCDC2 (P ¼ 0.01)(Supplementary outcome in prostate cancer patients treated with radical Table S1). When we confronted the data on immunohistochemical prostatectomy. Clinical follow-up data were available for patients staining for DCDC2 and ESE3/EHF in tissue microarray we found a in the two cohorts for which we had DCDC2 mRNA level similar inverse association between the two proteins. Interestingly, determined by microarray and qRT–PCR, respectively. In both 54% of prostate tumors with reduced ESE3/EHF expression (ESE3- cohorts we divided patients based on the DCDC2 mRNA level in Low) were positive for DCDC2. In contrast, 86% of ESE3 positive three groups: low (L) intermediate (I) and high (H) DCDC2 (ESE3-Pos) tumors were DCDC2 negative. These data suggested expressing tumors. Interestingly, Kaplan-Meier analysis using that expression of DCDC2 could be at least in part controlled by DCDC2 expression level from array data, indicated that the ESE3/EHF in prostate tumors. ESE3/EHF is an ETS transcription probability of biochemical recurrence-free survival in patients with factor with tumor-suppressor function and frequently lost in high level of DCDC2 was significantly higher than in patients with prostate tumors.12 Thus, we hypothesized that transcription of low and intermediate level of DCDC2 mRNA (Log-rank Mantel-Cox DCDC2 was negatively controlled by ESE3/EHF in normal prostate test ¼ 0.012) (Figure 1c). The association between increased level epithelial cells and reactivated as a consequence of ESE3/EHF of DCDC2 and biochemical relapse was also observed in the downregulation in cancer cells. cohort of FFPE prostate patients analyzed by qRT–PCR (Figure 1d), To further define the link between ESE3/EHF and DCDC2, we although the difference was not statistically significant (Log-rank determined whether there was a relationship between ESE3/EHF Mantel-Cox test ¼ 0.09). Collectively, these data show for the first and DCDC2 expression in prostate cancer cell lines. As shown in time the aberrant expression of DCDC2 in prostate tumors and its Figure 3a, DCDC2 was absent in ESE3/EHF positive LNCaP cells association with increased biochemical relapse and more aggres- although it was highly expressed in ESE3/EHF negative DU145 sive disease behavior. cells. To functionally link DCDC2 expression with ESE3/EHF loss in Expression of the neuronal-specific gene DCDC2 in prostate prostate cancer cells, we evaluated the expression of DCDC2 in tumors might be part of an altered differentiation program whose LNCaP cells with stable knockdown of ESE3/EHF (ESE3-KD).19 In various elements would contribute to tumorigenesis. To under- multiple ESE3-KD LNCaP clones, we observed a consistent increase stand whether DCDC2 expression was associated with altered of DCDC2 mRNA and protein (Figure 3b). Conversely, stable expression of other tumor promoting factors we performed a reexpression of ESE3/EHF in DU145 cells, reduced significantly the correlation analysis using the available prostate cancer microarray level of DCDC2 mRNA and protein compared with control cells dataset. We find 589 genes positively correlated (Pearson score (Figure 3c). 40 and P value o0.05) and 669 genes negatively correlated to Thus, our data indicated that the expression of ESE3/EHF and DCDC2 (Pearson score o0 and P value o0.05) (Supplementary DCDC2 was inversely correlated in human tumor samples and cell Table S1). Functional annotation analysis revealed that the genes lines, and suggested that ESE3/EHF could maintain the gene in a positively correlated with DCDC2 were functionally related to cell repressive state. We showed previously that ESE3/EHF could act as pathways linked to cell transformation and tumor progression a transcriptional repressor in addition to possessing transcrip- such as angiogenesis, signal transduction, cell migration and tional-activating functions depending on the promoter context.19

Oncogene (2013) 2315 – 2324 & 2013 Macmillan Publishers Limited DCDC2 is overexpressed in prostate cancer N Longoni et al 2317

Figure 1. DCDC2 mRNA is elevated in prostate cancers and associated with increased biochemical relapse. (a) DCDC2 expression in normal prostate (n ¼ 10) and prostate tumors (n ¼ 53) in microarray data. Left panel, data presented as log2 ratios relative to a common normal reference. Right panel, patient distribution based on DCDC2 RNA level from microarray data. (b) DCDC2 mRNA evaluated by QRT–PCR in archival formalin fixed paraffin embedded prostate cancer samples. Results are reported as DCDC2/b-actin ratio (left panel). Right panel, patient distribution based on DCDC2 RNA level determined by QRT–PCR. (c) Kaplan-Meier analysis of relapse-free survival of prostate cancer patients grouped according to the DCDC2 RNA level evaluated by microarray. Plot in the indicated DCDC2 expressing groups. (d) Kaplan- Meier analysis of relapse-free survival of prostate cancer patients grouped according to the DCDC2 RNA level determined by QRT–PCR. (e) Functional annotation analysis of the genes positively and negatively correlated with DCDC2 in microarray data.

To understand whether ESE3 directly controlled DCDC2, we to the DCDC2 promoter in cells not expressing DCDC2. These data, performed computational search of ETS binding sites in the along with the changes observed upon ESE3/EHF knockdown and DCDC2 promoter region. Using MotifViz, we identified numerous reexpression, indicate that ESE3/EHF has a repressive effect on putative ETS binding sites in the DCDC2 promoter with high score. DCDC2 transcription. To further confirm this point, we evaluated Specifically, our analysis identified a site located at –460/–471 with the effect of ESE3/EHF on DCDC2 promoter activity in luciferase the highest score that was selected for experimental validation reporter assays. Expression of ESE3/EHF significantly repressed the (Figure 3d). Chromatin immunoprecipitation showed binding of activity of the DCDC2-promoter reporter (Figure 3f). Collectively, ESE3/EHF in correspondence to this site in LNCaP cells (Figure 3e these data demonstrate that ESE3/EHF directly maintains the and Supplementary Figure 1). No binding was observed on the DCDC2 gene repressed in prostate epithelial cells and that loss of DCDC2 promoter in ESE3-KD LNCaP cells. Thus, ESE3/EHF is bound ESE3/EHF in tumors results in DCDC2 reexpression.

& 2013 Macmillan Publishers Limited Oncogene (2013) 2315 – 2324 DCDC2 is overexpressed in prostate cancer N Longoni et al 2318

Figure 2. Immunohistochemical evaluation of DCDC2 expression in normal prostate and prostate tumors. (a) Representative images of DCDC2 positive and negative prostate tissue samples. (b) Distribution of DCDC2 score intensity in normal prostate and primary prostate cancers in tissue microarrays. Number of samples in each class is indicated. (c) Percentage of DCDC2 positive and negative cases among ESE3 expressing (ESE3-POS) and nonexpressing (ESE3-LOW) prostate cancers. Number of samples in each class is indicated. (d) Representative images of tissue samples negatively and positively stained for DCDC2 and ESE3/EHF.

DCDC2 binds microtubules and promotes cell migration in DU145 cells, suggesting that DCDC2 expression did not affect prostate cancer cells a- expression and distribution. A similar pattern of DCX family proteins share the microtubule-binding DCX domain.11 intracellular distribution and colocalization with microtubules The DCX domain binds tubulin and enhances microtubule was seen in ESE3-KD LNCaP cells in which DCDC2 was reexpressed polymerization.15,18 We analyzed the subcellular localization of (Figure 4a, right panels). Instead, as in parental LNCaP cells, DCDC2 DCDC2 in prostate cancer cells by immunofluorescence microscopy. was undetectable in control (small hairpin RNA) shRNA transfected We were interested in determining DCDC2 localization and eventual LNCaP cells. Colocalization of DCDC2 and a-tubulin in DU145 cells association with microtubules when it is ectopically expressed in was confirmed by confocal microscopy (Figure 4b). To further prostate cancer cells. In DU145 cells, DCDC2 was present understand whether DCDC2 and a-tubulin directly interacted we predominantly in the cytoplasm with a ring-like distribution performed coimmunoprecipitation using DCDC2 and a-tubulin surrounding the nucleus (Figure 4, left panels), which is reminiscent antibodies. We observed that DCDC2 coimmunoprecipitated with of the pattern observed for DCX in primary cortical .17 a-tubulin (Figure 4c). Collectively, these data show for the first Staining with an a-tubulin antibody showed a large overlap time the DCDC2 intracellular localization in prostate cancer cells between DCDC2 and microtubules. The DCDC2 appearance was and indicate a direct interaction of the protein with microtubules punctuated than the a-tubulin immunostaining. However, the also in the prostate-cell context. merge revealed significant areas of colocalization of the two Interaction between the DCX domain and microtubules has proteins. The regions of overlap between DCDC2 and a-tubulin been associated to the migratory activity in brain . immunoreactivity were mostly surrounding the nucleus. This pattern Consistently, it has been reported that local loss of function of of colocalization was consistent with that observed with DCX and DCDC2 induced by RNA interference result in the interruption of other microtubules-associated protein (MAP) like tau.20–22 normal neuronal migration.13 Thus, we hypothesize that Consistent with the ESE3/EHF repressive effect on DCDC2 reexpression of DCDC2 in prostate cells could alter the cell expression, there was very little staining for DCDC2 in DU145 cells migratory capability. To test this, we used RNAi to knockdown stably expressing ESE3/EHF (Figure 4a). Interestingly, a-tubulin DCDC2 in DU145 and ESE3-KD LNCaP cells. Successful DCDC2 immunostaining was similar in DCDC2 positive and negative knockdown was documented at the RNA level (Figure 5a).

Oncogene (2013) 2315 – 2324 & 2013 Macmillan Publishers Limited DCDC2 is overexpressed in prostate cancer N Longoni et al 2319

Figure 3. DCDC2 is inversely correlated with ESE3/EHF in prostate cancer cell lines. (a) ESE3 and DCDC2 expression in LNCaP and DU145 cells determined by RT–PCR (left panel) and immunoblotting (right panel). (b) DCDC2 expression in control (sh-) and ESE3 knockdown (sh 3, 6 and 7) LNCaP clones determined by RT–PCR (left panel) and immunoblotting (right panel). (c) DCDC2 expression in DU145 cells stably transfected with an empty (pcDNA) or ESE3/EHF expression vector (pESE3) determined by RT–PCR (left panel) and immunoblotting (right panel). (d) Position of the ETS binding site (arrow, at nt À 336–327) identified with MotifViz in the DCDC2 promoter region probed in ChIP assays. (e) Binding of ESE3/ EHF to the DCDC2 promoter determined by ChIP and qPCR in control (sh-) and ESE3 knockdown (sh7) LNCaP cells. (f) Luciferase activity of the DCDC2 promoter reporter transfected in DU145 cells along with either pcDNA or ESE3/EHF expression vector (pESE3). **Po0.01.

Knockdown of DCDC2 significantly reduced cell migration in the DCDC2 is involved in taxol resistance wound healing assay in both cell types (Figure 5b). To rule out DCX proteins are MAPs and data showed above indicate that target off effects of the small interfering RNA (siRNA), we next DCDC2 colocalizes with microtubules in prostate cancer cells. targeted DCDC2 with shRNA. Knockdown of DCDC2 by two Microtubules are essential components of the and shRNAs (sh1 and sh3) targeting different regions of the gene have a critical role in many cellular process including cell motility resulted in significant DCDC2 downregulation compared with and shape maintenance.23,24 Their dynamic properties are crucial control shRNA (sh-) (Supplementary Figure 2A) and resulted in a for the assembly of the mitotic spindle and attachment of significantly reduced cell migration as assessed by wound healing along the spindle. Microtubule-targeting drugs, (Figure 5c and Supplementary Figure 2B) and Boyden chamber such as taxol, bind to tubulin and enhance microtubule assay in DU145 (Figure 5d). In contrast, knockdown of DCDC2 did polymerization.23,24 MAP, by binding to tubulin, can alter not affect cell proliferation (Supplementary Figure 2C). Collec- microtubule dynamics and thus the sensitivity of cells to tively, these results indicate that DCDC2 has a direct role in cell microtubule-targeting drugs.23,24 Indeed, expression of various migration and that reexpression of the gene following ESE3/EHF MAPs has been linked to the level of sensitivity of cancer cells to loss is associated with the acquisition of a migratory phenotype in microtubule-targeting drugs.22,25,26 High levels of tau, a MAP prostate epithelial cells. protein related with DCDC2, have been associated with resistance

& 2013 Macmillan Publishers Limited Oncogene (2013) 2315 – 2324 DCDC2 is overexpressed in prostate cancer N Longoni et al 2320

Figure 4. Intracellular localization of DCDC2 in prostate cancer cells and DCDC2–a-tubulin interaction. (a) Left panels, Immunofluorescence staining for DCDC2 and a-tubulin in control (pcDNA) and ESE3- expressing (pESE3) DU145 cells. Right panels, Immunofluorescence staining of DCDC2 and a-tubulin in control (sh-) and ESE3 knockdown (sh7) LNCaP cells. (b) Immunofluorescence staining for DCDC2 and a-tubulin in DU145 cells by confocal microscopy in DU145 cells (c) DCDC2 co-immunoprecipitates with a-tubulin in DU145 cells. Cellular extract were immunoprecipitated with DCDC2 and a-tubulin antibodies or with immunoglobulin 1(IgG) as control. Western-blots were then performed with antibodies against DCDC2 and a-tubulin.

to taxol in breast cancers22 and loss of tau expression sensitized wondered whether reexpression of DCDC2 in ESE3-KD cells could breast cancer cells to taxol, consistent with its function in contribute to the induction of taxol resistance. microtubule stabilization.22 Therefore, we hypothesized that To determine the role of DCDC2, we knocked down DCDC2 in DCDC2 could alter the response of prostate cancer cells to taxol ESE3-KD LNCaP cells and evaluated the effects of taxol in by virtue of its ability to interact with microtubules. In keeping cytotoxicity assays. Parental LNCaP cells were highly sensitive to with this hypothesis, we observed that the apoptotic response to taxol and transfection with a siRNA specific for DCDC2 did not taxol, measured by PARP cleavage, was active in LNCaP cells affect their response to the treatment (Figure 6c). In contrast, we expressing ESE3/EHF, although it was greatly attenuated in ESE3- observed a significant decrease in the percentage of viable cells KD LNCaP cells (Figure 6a). Furthermore, cytotoxicity assays upon knockdown of DCDC2 in ESE3-KD LNCaP cells, which showed that ESE3-KD LNCaP cells were more resistant than overexpress DCDC2. Consistently, reexpression of DCDC2 by parental LNCaP cells to taxol-induced cell death (Figure 6b). This transfection of an expression vector in LNCaP cells resulted in effect was clearly linked to ESE3/EHF expression level with loss of increased resistance to taxol (Figure 6d). Furthermore, knockdown ESE3, resulting in increased resistance to taxol. Therefore, we of DCDC2 in DU145 cells resulted in a significant increased

Oncogene (2013) 2315 – 2324 & 2013 Macmillan Publishers Limited DCDC2 is overexpressed in prostate cancer N Longoni et al 2321

Figure 5. DCDC2 increases migration of prostate cancer cells. (a) DCDC2 knockdown assessed by RT-PCR in DU145 and LNCaP cells. (b) Wound healing assay in DU145 and LNCaP cells transfected with control (siGL3) or DCDC2 directed siRNA. Pictures were taken at the indicated times. Representative photographs of triplicate experiments are shown. (c) Quantification of wound healing assay data from triplicate experiments. Data represent mean±s.d. of the percentage of wound width relative to time 0. **Po0.01. response to taxol (Figure 6e and Supplementary Figure 3). microtubules. Mutations in the human DCX gene causes double Collectively, these studies identify a novel mechanism of taxol cortex syndrome and lissencephaly in humans.27 In both diseases resistance in prostate tumors mediated by the aberrant expression the point mutations cluster within the conserved DCX peptide of DCDC2. motifs of DCX that serve as protein-interaction platforms and interaction with microtubules. Here we report that DCDC2 is ectopically expressed in about DISCUSSION 50% of prostate tumors, whereas it is absent in normal prostate. In this study, we show that the gene encoding DCDC2 is a novel By integrating gene expression profile and tissue microarray data oncogenic target activated in prostate tumors. DCDC2 is a in human tumors with experiments in human cancer cell lines, we member of the DCX protein family, which is characterized by show that ectopic expression of this neuronal protein in prostate the presence of DCX peptide domains.17 The DCX gene, located tumors is associated with oncogenic transformation and tumor on the X and the first characterized gene of the progression. Our data indicate that in prostate tumors high family, encodes a cytoplasmic protein that directs neuronal expression of DCDC2 is associated with worse prognosis and migration by regulating the organization and stability of significantly shorter relapse-free survival. Interestingly, DCX has

& 2013 Macmillan Publishers Limited Oncogene (2013) 2315 – 2324 DCDC2 is overexpressed in prostate cancer N Longoni et al 2322 been proposed as a molecular marker to detect minimal residual the mechanism leading to DCDC2 activation in prostate disease in neuroblastoma.28,29 Moreover, DCX was reported to be cancer. We showed previously that the ETS transcription factor preferentially expressed in invasive human brain tumors.30 Thus, ESE3/EHF is present in normal prostate and downregulated in our results are consistent with the notion that aberrant expression prostate tumors.10 Through analysis of microarray and of DCX family proteins in human tumors may have clinical immunohistochemical data, we found that the DCDC2 expression relevance and suggest that a member of this family, DCDC2, may was inversely correlated with ESE3/EHF. A similar inverse be particularly relevant in the context of prostate tumors. This correlation was found in prostate cancer cell lines supporting a finding deserves further investigation as expression of DCDC2 may link between reduced expression of ESE3/EHF and activation of mark tumors with a higher degree of dedifferentiation and DCDC2 transcription. Furthermore, knockdown of ESE3/EHF led to particularly aggressive features. expression of DCDC2, although re-expression of ESE3/EHF in The restricted pattern of expression of DCDC2 in prostate tumors cancer cells reduced the level of DCDC2. Consistently, we identified and its almost complete absence in normal tissue suggested an ETS binding site in the DCDC2 promoter and confirmed by a close link to oncogenic transformation. We investigated chromatin immunoprecipitation the binding of ESE3/EHF to the DCDC2 promoter in cells that do not express DCDC2. Thus, the epithelial-specific ETS factor ESE3/EHF maintains the DCDC2 gene repressed, preventing its aberrant expression in prostate epithelial cells. These results uncover also an important mechanism by which ESE3/EHF could exert its tumor-suppressor function, that is, by preventing the expression of proteins contributing to the acquisition of migratory capability and metastatic phenotype. Indeed, functional experiments showed that DCDC2 expression increased in vitro cell migration. Intriguingly, although the effect on cell migration was consistent with the role of DCDC2 in neuronal cells, this work identifies an additional and potentially clinically relevant consequence of aberrant expression of DCDC2 in prostate tumors. We found that DCDC2 expression was associated with increased resistance to taxol in prostate cancer cells. This effect is likely linked to the ability of the DCDC2 to interact with microtubules and interfere with taxol-induced disruption of microtubule function and the consequent activation of the apoptotic cascade. Members of the DCX protein family bind to microtubules through their conserved DCX motifs, promoting microtubule assembly and stabilization.15,17,18,31 We observed colocalization of DCDC2 with microtubules in prostate cancer cells, and direct interaction of DCDC2 with a-tubulin, suggesting that DCDC2 could act through a similar mechanism of microtubule stabilization to alter the sensitivity of prostate cancer cells to taxol. Altered expression of other MAPs, like tau, has been shown to be associated with taxol resistance in breast cancer.22 DCDC2 could have a similar effect when overexpressed in prostate tumors and mark tumors potentially less responsive to taxol treatment. Overall, the present data highlight a novel ETS-related oncogenic target in prostate cancer. We found that DCDC2 is involved in cell migration and taxol resistance in prostate cancer cells and is negatively associated with disease recurrence. Future studies are warranted to explore the clinical relevance of DCDC2

Figure 6. DCDC2 expression promotes taxol resistance in prostate cancer cells. (a) Representative immunoblotting showing PARP cleavage in control (sh-) and ESE3 knockdown (sh7) LNCaP cells following a 24-h treatment with taxol at the indicated doses. a-tubulin was used as loading control. (b) Viability of control (sh-) and ESE3 knockdown (sh7, sh14) LNCaP cells following 24-h treatment with taxol. Cell viability was assessed by MTT assay after 72 h. (c) Knockdown of DCDC2 restores sensitivity of ESE3 knock- down LNCaP cells to taxol. Control (sh-) and ESE3 knockdown (sh7) LNCaP cells were transfected with control (siGL3) or DCDC2 directed siRNA and treated with the indicated concentration of taxol for 24 h. Cell viability was assessed by MTT assay after 72 h. (d) Expression of DCDC2 in LNCaP cells increase resistance to taxol. LNCaP cells were transfected with DCDC2 full-length expression vector (pDCDC2) or with control pcDNA. After 24 h, cells were treated with the indicated concentration of taxol for 24 h. Cell viability was assessed by MTT assay after 48 h. (e) Knockdown of DCDC2 increases sensitivity of DU145 cells to taxol. DU145 cells were transfected with control (siGL3) or DCDC2 directed siRNA and treated with the indicated concentration of taxol for 24 h. Cell viability was assessed by after 72 h. *Po0.05; **Po0.01.

Oncogene (2013) 2315 – 2324 & 2013 Macmillan Publishers Limited DCDC2 is overexpressed in prostate cancer N Longoni et al 2323 as a diagnostic, prognostic and predictive marker of therapeutic Corporation, Fremont, CA, USA) followed by incubation with a secondary response to taxanes. mouse anti-rat antibody (DAKO). Samples, in which the primary antibody was omitted, were included as negative controls. Staining was independently scored by three expert pathologists. MATERIALS AND METHODS Microarray data analysis Cell culture, transfection and drug treatment Gene expression profiling of normal prostate and tumor samples has been 10 LNCaP and DU145 were obtained from American Type Culture Collection previously described and patient characteristics reported. Gene and maintained in RPMI-1640 supplemented with 10% fetal bovine expression analysis was performed on Agilent Human 1A glass arrays serum as previously described.19 ESE3/EHF expressing DU145 cells were using a dye-swap replication scheme. Data are MIAME compliant and have generated after transfection with ESE3/EHF expression vector and selected been deposited in the Gene Expression Omnibus (GEO accession number. with G418.10 Negative control cells were obtained by transfection with GSE14206). A commercial pool of RNA from organ donor healthy prostates pcDNA3.1 (Invitrogen, Zug, Switzerland) and selection in G418. LNCaP (Becton Dickinson, Buccinasco, Milano, Italy) was used as common stably expressing ESE3 targeting shRNAs were established as previously reference. To identify differentially expressed genes between normal and described.19 For transient gene knockdown, cells were transfected with tumor samples a log2 gene expression matrix was created after combining siRNAs targeting DCDC2 or control siRNA using Interferin.10 Alternatively, dye-swap replicates. Expression data were then filtered for s.d.40.5 across sh plasmid containing siRNA against DCDC2 or control sh (SABiosciences, the samples, giving a matrix containing 5142 probes. Differential gene QIAGEN AG, Hombrechtikon, Switzerland) were transfected in DU145 cells expression analysis between sample classes was performed on the 5142- with JetPrime reagent (Polyplus-transfection SA, Illkirch, France), according gene matrix using Gene Expression Profile Analysis Suite. Differentially to manufacturers protocol. For DCDC2 overexpression DCDC2-V5 plasmid expressed genes were obtained after filtering for q-value (Qp0.1). containing full length DCDC2-GFP tagged (kindly provided by Dr Satu To identify genes correlated and anticorrelated with DCDC2, we used Massinen33) and pcDNA were transfected using JetPrime (Polyplus- the R function correlation test (), which provides Pearson’s correlation Transfection SA, Illkirch, France). Transfection efficiency was assessed coefficient. evaluating GFP positive cells by microscopy and Western-blot (data not shown). Cell proliferation was assessed 48 and 72 h after transfection using Patient samples a colorimetric assay (MTT, Sigma) and reading absorbance at 540 nM in a 34 Tissue samples from the FFPE tissue cohort used for qRT–PCR validation of microplate reader. Taxol treatment was performed in 96-well plates and DCDC2 were obtained from patients with organ-confined prostate cancer cell viability measured with MTT as described above. All assays were done treated with radical prostatectomy at Hospital of Biella (Italy), between in triplicate and repeated in three independent experiments. 1993 and 1998, and collected with the approval of the Ethical Committee of the Piedmont Region and patient’s written informed consent. Samples for immunohistochemical studies were obtained from surgically Luciferase promoter assay 19 treated patients with organ-confined prostate cancer at the IRCCS (Italian Luciferase assays were carried out as previously described using a Research Hospital) Multimedica, Italy, with the approval of the Regione luciferase promoter reporter containing 600 base pairs of the DCDC2 35 Lombardia Ethical Committee and patient’s written informed consent. promoter region (kindly provided by Dr Jeffrey Gruen).

RNA extraction from tumor samples and qRT–PCR Cell migration RNA was extracted from tissue sections from archival FFPE samples. Two To assess cell migration in wound healing/scratch assays cells were grown expert pathologists (FS and GD) selected the area to be processed. Briefly, to confluence in six-well plates and then overnight in serum-free medium. for every tissue-block a slice was stained with H&E and evaluated to Then, scratches were performed on the cell monolayer, complete medium identify areas enriched of tumoral lesions (at least 75% of tumor cells). was added and cells monitored for 24–48 h using a Zeiss Axiovert Subsequent slices were used to isolate RNA from macrodissected areas microscope equipped with Axiocam Mrc3 (Carl Zeiss AG, Feldbach, based on the H&E staining. Tissue were deparaffinized with ethanol and Switzerland). For quantitative analysis wound size was measured at each RNA was extracted using RNeasy Mini Kit (Qiagen, Hombrechtikon, time point in triplicate wells and data represent the average of multiple Switzerland) and then treated with DNase. QRT-PCR was performed using experiments. For the Boyden chamber assay, 24-well Insert System with an SYBR Green qPCR. Primer sequences are shown in supplementary Table S2. 8 mM PET membrane (BD Bioscience, Allschwil, Switzerland) was coated with Collagen type IV (Sigma-Aldrich Chemie GmbH, Buchs, Switzerland). Survival analysis Cells were added to the top of chamber in RPMI with 0.1% BSA. Using Kaplan-Meier survival curves were created using SPSS software version DU145 conditioned media as chemoattractant, cells were incubated at 1 17.0 (IBM SPSS, Chicago, IL, USA). The log rank test was applied to examine 37 C for 24 h. After incubation, nonmigrating cells were carefully removed the relationship between DCDC2 expression and biochemical progression- from the top of each insert with a cotton swab. Migrating cells were fixed, free survival. The level of DCDC2 in the microarray data set was defined as stained and counted microsopically using Zeiss Axiovert (X100 objective) to determine the extent of cell migration. The assays were done in low (L), Intermediate (I) and High (H), whether it was, respectively, o2, X2 and 4threefold different from the value in normal prostate samples. triplicate and repeated in three independent experiments. Biochemical recurrence was defined as a 0.2 ng/ml increase in prostate specific antigen with a second confirmatory prostate specific antigen RT–PCR and immunoblotting measurement 40.2 ng/ml. Patients were censored at the time of the last clinical follow-up. For the microarray data set,19 patients included in the RNA was extracted from cell lines using Trizol (Invitrogen) and treated with DNase I to remove any contaminant genomic DNA. Conventional real-time analysis (n ¼ 28) had completed eight years of follow-up, had been treated 10 with radical prostatectomy and had not received any treatment before and PCR (RT–PCR) was performed as described previously using SuperScript following surgery until the time of relapse. For the archival patients FFPE One-step RT–PCR System (Invitrogen). PCR products were analyzed by gel data set, the same criteria were applied and patients included in the electrophoresis and visualized using the AlphaImager 3400 (Proteinsimple, Santa Clara, CA, USA). Cell lysate preparation, gel electrophoresis analysis (n ¼ 27) had completed 13 years of follow-up. Patient’s analysis is 10 summarized in Supplementary Figure 4. and immunoblotting were performed as described previously using antibodies against DCDC2 (Santa Cruz Biotechnology Inc., Heidelberg, Germany), ESE3/EHF (Clone 5A5.5, Lab Vision, Fremont CA, USA), cleaved Immunohistochemistry PARP (BD Pharmingen, Switzerland), and a-tubulin (Santa Cruz Tissue microarrays containing samples from normal and adenocarcinoma Biotechnology Inc). For immunoprecipitation, DU145 and LNCaP cells 32 tissues were prepared from FFPE tissues as previously described. Sections were lysed in RIPA Buffer (50 mM Tris-HCl ph 7.4, 150 mM NaCl, 1 mM EDTA, were dewaxed in xylol, rehydrated in ethanol and treated with antigen 1% NP40, 0.25% Na-deocycholate, 1 mM PMSF, 1 mM NaF, 1X (Roche, Basel, retrieval solution (DAKO, Glostrup, Denmark) as recommended by the Switzerland) complete mini protease inhibitor cocktail). Protein extracts manufacturer. Immunohistochemistry was carried out using anti-DCDC2 (200 mg) were incubated with anti-DCDC2 and anti-a-tubulin antibodies rabbit polyclonal antibody (1:200 dilution, Sigma, Buchs, Switzerland) and overnight at 4 1C and then incubated with protein G agarose beads for 4 h. anti-ESE-3 (Clone 5A5.5) rat monoclonal antibody (1:75 dilution, LabVision The beads were washed with RIPA buffer four times and then eluted

& 2013 Macmillan Publishers Limited Oncogene (2013) 2315 – 2324 DCDC2 is overexpressed in prostate cancer N Longoni et al 2324 in 40 ml of loading buffer. The immunoprecipitates were analyzed as 14 Koizumi H, Higginbotham H, Poon T, Tanaka T, Brinkman BC, Gleeson JG et al. 10 described previously using antibodies against DCDC2 and a-tubulin. Doublecortin maintains bipolar shape and nuclear translocation during migration in the adult forebrain. Nat Neurosci 2006; 9: 779–786. Immunofluorescence and fluorescence microscopy 15 Gleeson JG, Lin PT, Flanagan LA, Walsh CA. Doublecortin is a microtubule- Cells were grown on glass coverslips, fixed with 4% formaldehyde, associated protein and is expressed widely by migrating neurons. Neuron 1999; permeabilized with acetone, stained with FITC-labeled phalloidin (Invitro- 23: 257–271. gen) and incubated with anti-DCDC2 and anti-a-Tubulin and then with 16 Burbridge TJ, Wang Y, Volz AJ, Peschansky VJ, Lisann L, Galaburda AM et al. anti-rabbit Alexa 488 or anti-mouse Alexa 594 (Invitrogen). Pictures were Postnatal analysis of the effect of embryonic knockdown and overexpression of taken with Zeiss Axiovert (100X objective) using Axiocam Mrc3 and candidate susceptibility gene homolog Dcdc2 in the rat. Neuroscience processing with AxioVision 4.5 (Carl Zeiss AG). 2008; 152: 723–733. 17 Coquelle FM, Levy T, Bergmann S, Wolf SG, Bar-El D, Sapir T et al. Common and Chromatin Immunoprecipitation divergent roles for members of the mouse DCX superfamily. Cell Cycle 2006; Cells were processed using the ChIP assay kit from Upstate Biotechnology 5: 976–983. (Walthman, MA, USA) according to the manufacturer’s protocol as 18 Horesh D, Sapir T, Francis F, Wolf SG, Caspi M, Elbaum M et al. Doublecortin, a previously described.10 Chromatin was immunoprecipitated with an anti- stabilizer of microtubules. Hum Mol Genet 1999; 8: 1599–1610. ESE3 antibody (Clone 5A5.5, Lab Vision, Fremont, CA, USA). qPCR was 19 Kunderfranco P, Mello-Grand M, Cangemi R, Pellini S, Mensah A, Albertini V et al. performed as described using primers specific for the gene of interest (see ETS transcription factors control transcription of EZH2 and epigenetic silencing supplementary Table S2). Each experiment was done in triplicate and of the tumor suppressor gene Nkx3.1 in prostate cancer. PLoS One 2010; 5: repeated at least three times. Representative results are shown. e10547. 20 Tint I, Jean D, Baas PW, Black MM. Doublecortin associates with microtubules preferentially in regions of the axon displaying actin-rich protrusive structures. CONFLICT OF INTEREST J Neurosci 2009; 29: 10995–11010. The authors declare no conflict of interest. 21 Dijkmans TF, van Hooijdonk LW, Fitzsimons CP, Vreugdenhil E. The doublecortin gene family and disorders of neuronal structure. Cent Nerv Syst Agents Med Chem 2010; 10: 32–46. ACKNOWLEDGEMENTS 22 Rouzier R, Rajan R, Wagner P, Hess KR, Gold DL, Stec J et al. Microtubule-asso- This work was supported by grants from Oncosuisse (KFS-01913-08 and KFS-02573- ciated protein tau: a marker of paclitaxel sensitivity in breast cancer. Proc Natl 02-2010), Swiss National Science Foundation (FNS-31003A-118113) and Ticino Acad Sci USA 2005; 102: 8315–8320. Foundation for Cancer Research to GMC. MMG and GC were supported by 23 Dumontet C, Jordan MA. Microtubule-binding agents: a dynamic field of cancer Compagnia di San Paolo, Torino, Italy. therapeutics. Nat Rev Drug Discov 2010; 9: 790–803. 24 Mollinedo F, Gajate C. Microtubules, microtubule-interfering agents and apop- tosis. Apoptosis 2003; 8: 413–450. REFERENCES 25 Wagner P, Wang B, Clark E, Lee H, Rouzier R, Pusztai L. Microtubule Associated Protein (MAP)-Tau: a novel mediator of paclitaxel sensitivity in vitro and in vivo. 1 DeMarzo AM, DeMarzo AM, Nelson WG, Isaacs WB, Epstein JI. Pathological and molecular aspects of prostate cancer. Lancet 2003; 361: 955–964. Cell Cycle 2005; 4: 1149–1152. 26 Soltani MH, Pichardo R, Song Z, Sangha N, Camacho F, Satyamoorthy K et al. 2 Seth A, Watson DK. ETS transcription factors and their emerging roles in human cancer. Eur J Cancer 2005; 41: 2462–2478. Microtubule-associated protein 2, a marker of neuronal differentiation, induces 3 Tomlins SA, Rhodes DR, Perner S, Dhanasekaran SM, Mehra R, Sun X et al. mitotic defects, inhibits growth of melanoma cells, and predicts metastatic Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate potential of cutaneous melanoma. Am J Pathol 2005; 166: 1841–1850. cancer. Science 2005; 310: 644–648. 27 Dobyns WB, Truwit CL, Ross ME, Matsumoto N, Pilz DT, Ledbetter DH et al. 4 Kumar-Sinha C, Tomlins SA, Chinnaiyan AM. Recurrent gene fusions in prostate Differences in the gyral pattern distinguish chromosome 17-linked and X-linked cancer. Nat Rev Cancer 2008; 8: 497–511. lissencephaly. Neurology 1999; 53: 270–277. 5 Klezovitch O, Risk M, Coleman I, Lucas JM, Null M, True LD et al. A causal role for 28 Oltra S, Martinez F, Orellana C, Grau E, Fernandez JM, Can˜ete A et al. The dou- ERG in neoplastic transformation of prostate epithelium. Proc Natl Acad Sci USA blecortin gene, a new molecular marker to detect minimal residual disease in 2008; 105: 2105–2110. neuroblastoma. Diagn Mol Pathol 2005; 14: 53–57. 6 Tomlins SA, Laxman B, Varambally S, Cao X, Yu J, Helgeson BE et al. Role of the 29 Viprey VF, Lastowska MA, Corrias MV, Swerts K, Jackson MS, Burchill SA. Minimal TMPRSS2-ERG gene fusion in prostate cancer. Neoplasia 2008; 10: 177–188. disease monitoring by QRT-PCR: guidelines for identification and systematic 7 Carver BS, Tran J, Gopalan A, Chen Z, Shaikh S, Carracedo A et al. Aberrant ERG validation of molecular markers prior to evaluation in prospective clinical trials. expression cooperates with loss of PTEN to promote cancer progression in the J Pathol 2008; 216: 245–252. prostate. Nat Genet 2009; 41: 619–624. 30 Daou MC, Smith TW, Litofsky NS, Hsieh CC, Ross AH. Doublecortin is preferentially 8 King JC, Xu J, Wongvipat J, Hieronymus H, Carver BS, Leung DH et al. expressed in invasive human brain tumors. Acta Neuropathol 2005; 110: 472–480. Cooperativity of TMPRSS2-ERG with PI3-kinase pathway activation in prostate 31 Sapir T, Horesh D, Caspi M, Atlas R, Burgess HA, Wolf SG et al. Doublecortin oncogenesis. Nat Genet 2009; 41: 524–526. mutations cluster in evolutionarily conserved functional domains. Hum Mol Genet 9 Zong Y, Xin L, Goldstein AS, Lawson DA, Teitell MA, Witte ON. ETS family 2000; 9: 703–712. transcription factors collaborate with alternative signaling pathways to induce 32 Bowen C, Bubendorf L, Voeller HJ, Slack R, Willi N, Sauter G et al. Loss of NKX3.1 carcinoma from adult murine prostate cells. Proc Natl Acad Sci USA 2009; 106: expression in human prostate cancers correlates with tumor progression. Cancer 12465–12470. Res 2000; 60: 6111–6115. 10 Cangemi R, Mensah A, Albertini V, Jain A, Mello-Grand M, Chiorino G et al. 33 Massinen S, Hokkanen ME, Matsson H, Tammimies K, Tapia-Pa´ez I, Dahlstro¨m- Reduced expression and tumor suppressor function of the ETS transcription factor Heuser V et al. Increased expression of the dyslexia candidate gene DCDC2 affects ESE-3 in prostate cancer. Oncogene 2008; 27: 2877–2885. length and signaling of primary cilia in neurons. PLoS One 6: e20580. 11 Reiner O, Coquelle FM, Peter B, Levy T, Kaplan A, Sapir T et al. The evolving 34 Albertini V, Jain A, Vignati S, Napoli S, Rinaldi A, Kwee I et al. Novel GC-rich doublecortin (DCX) superfamily. BMC Genomics 2006; 7: 188. DNA-binding compound produced by a genetically engineered mutant of the 12 Kim MH, Cierpicki T, Derewenda U, Krowarsch D, Feng Y, Devedjiev Y et al. mithramycin producer Streptomyces argillaceus exhibits improved transcriptional The DCX-domain tandems of doublecortin and doublecortin-like kinase. Nat repressor activity: implications for cancer therapy. Nucleic Acids Res 2006; 34: Struct Biol 2003; 10: 324–333. 1721–1734. 13 Meng HY, Smith SD, Hager K, Held M, Liu J, Olson RK et al. DCDC2 is associated 35 Meng H, Powers NR, Tang L, Cope NA, Zhang PX, Fuleihan R et al. A dyslexia- with and modulates neuronal development in the brain. associated variant in DCDC2 changes gene expression. Behav Genet 2011; 41: Proc Natl Acad Sci USA 2005; 102: 17053–17058. 58–66.

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

Oncogene (2013) 2315 – 2324 & 2013 Macmillan Publishers Limited