Abnormal Expression of the ERG Transcription Factor in Prostate Cancer Cells Activates Osteopontin Free

Prostate cancer (PCa) is the most commonly diagnosed nonskin cancer in men and continues to be one of leading causes of male cancer-related death among the elderly. PCa has high metastatic capacity and extensively metastasizes to bone, lymph nodes, and visceral organs, such as the liver and lungs. Research over the last few decades on metastasis has revealed detailed steps of this mechanistic cascade (1): some primary tumor cells detach and escape from the original tumor sites, intravasate into circulation, extravasate out into the secondary tissues in which they begin to proliferate (2). Chemotaxis plays an important role in the metastasis of cancer cells, directing the motility of metastatic cells through gradients of growth factors or chemoattractants.

Osteopontin (OPN), also called secreted phosphoprotein 1 (SPP1), is a glycophosphoprotein cytokine that is expressed by numerous cells and secreted into body fluids. This extracellular matrix protein is an active player in many physiological and pathological processes including bone mineralization, defense against infectious agents, blood vessel formation, arteriosclerosis, disruption of the growth of calcium oxalate crystals and nitric oxide production, and tumorigenesis (3, 4). Substantial evidence associates OPN expression with tumor invasion and metastasis (5) in a number of cancers such as breast (6), prostate (7), lung (8), stomach (9) colon (10), brain (11), and other cancers (12).

Transcriptional regulation of OPN is complex and involves multiple signal transductions (13). Identifying transcriptional regulators that contribute to the modulation of OPN expression is of interest for therapy that aims to control the OPN-mediated metastatic phenotype. Several known cis-acting transcription factors have been described and most of them have been localized to a conserved region at 250 bp upstream of the proximal promoter. Potential binding sites for transcriptional regulators, such as AP-1, Myc, Oct-1, USF, v-Src, TGF-b/BMPs/Smad/Hox, Wnt/β-catenin/APC/GSK-3b/Tcf-4, Ras/RRF, TF53, Runx2, and ETS family members, have been identified (14–21).

In the past few years, the ETS-related gene (ERG gene), a member of the ETS transcriptional factor family, has been shown to be highly overexpressed in most PCa (60%–80% according to the studies; refs 22, 23). Thus, its presence in prostate cells perturbs normal gene expression. ERG overexpression (and to a lesser extent the overexpression of some other ETS family members) is the result of chromosomal translocations between the androgen-regulated Trans-membrane protease serine 2 (TMPRSS2) gene promoter and the coding sequence of the ERG gene (24). The TMPRSS2:ERG fusion is associated with aggressiveness and recurrence of PCa (25), metastasis (26) and PCa-specific death (27, 28).

Two members of the ETS family, PEA3 and Ets-1, have been shown to be partially responsible for the overexpression of OPN in human breast cancer and in murine colorectal cancer cells, respectively (29, 30). In this study, we investigated whether ERG overexpression stimulates OPN expression in PCa cells. OPN promoter–luciferase deletion constructs in transient transfection experiments showed OPN transactivation by the ERG factor. After localizing ETS binding sites (EBS) in sequences conserved across several species, we characterized the OPN promoter–ERG transcription factor complex using an electromobility shift assay (EMSA) and chromatin immunoprecipitation (ChIP). Because of the expression of the TMPRSS:ERG fusion gene in prostate tumor cells, the overexpression of ERG in PCa cells increased OPN expression. The coincident expression and colocalization of TMPRSS2:ERG and OPN in primary prostate tumor tissue samples suggest that ERG may be implicated in the dysregulation of OPN in PCa.

All PCa patients included in this study had undergone radical prostatectomy in the Lille University Hospitals. Clinical data and patient consent were provided by the referring physician. Immunohistochemical studies were conducted on formalin-fixed and paraffin-embedded prostate specimens. Frozen tissues used for RT-PCR involved primary tumors from 35 male patients (mean age: 63 years, with Gleason score >5) and were obtained from the urological collection of the local tumor tissue bank (Tumorothèque CRRC/Canceropole Nord-Ouest, Lille, France) after approval by the internal review board.

We used the expression plasmids for ERG (pSG5-ERGp55 and pcDNA3.1-ERG), described in reference (31). The TMPRSS2 (exon 1):ERG (exon 4) fusion transcript was amplified and subcloned into pcDNA3.1[(+) Invitrogen] using patient cDNA samples and primers: TMPRSS2 forward primer 5′-CGCGAGCTAAGCAGGAGGC-3′ and ERG reverse primer 5′-CCTCCGCCAGGTCTTTAGTA-3′. OPN promoter luciferase reporters were constructed by PCR with the following primers: OPN-116/+77 (−116 to +77), OPN-136/+77 (−136 to +77), OPN-341/+77 (−341 to +77), and OPN-1441/+77 (−1441 to +77). The fragments were subsequently cloned into a pGL3-basic luciferase reporter plasmid. Consensus ERG binding site mutations corresponding to nt −121 to −114 (5′-GGAGGAAG-3′ to 5′-GGTAAAAG-3′) and to nt −112 to −106 (5′-GTAGGAG-3′ to 5′-GTCGGAG-3′) were constructed using the OPN-136/+77 luciferase wild-type fragment as the template. Oligonucleotides encoding for small hairpin RNAs (shRNA) against ERG mRNA [targeted sequence described in (32)] was synthesized with appropriate loop and cohesive ends sequences according to the plasmid provider instructions and was cloned into pSilencer 2.1-U6 (Ambion Inc). All constructs were verified by sequencing (Genoscreen) and restriction enzyme digestion.

PC3c (a PC3 subclone provided by E.B.) and HeLa cell lines were grown in Dulbecco's modified Eagle's medium (DMEM; Gibco) supplemented with 10% fetal bovine serum, 1% gentamicin and 1% glutamin (Invitrogen). Cultures were maintained at 37°C under 5% CO₂.

One day prior to transfection, cells were plated in 12-well plates to be 50% to 60% confluent on the day of transfection. Each well was transfected using polyethylenimine (PEI, Eurogentec) according to the manufacturer's instructions, with a DNA mixture including 500 ng of the firefly luciferase reporter gene containing the OPN promoter, 25 ng of a control plasmid tk-luciferase (tk-renilla; Promega), and the indicated concentration of pSG5-ERG. In the RNA silencing analysis, 300 ng of pSilencer ERG or pSilencer control were cotransfected. The total amount of transfected DNA was adjusted to 1 μg per well. After 48 hours of transfection, cells were harvested and luciferase activities were assayed using the Dual-Luciferase Reporter Assay System (Promega) using a luminometer (Berthold Biolumat centro LB960). Firefly luciferase values were normalized to those of the control renilla luciferase values. All experiments were repeated twice. Results are presented as the mean ± SEM.

PC3c cells stably expressing ERG or TMPRSS2:ERG were obtained by transfecting them with pcDNA3-ERG or pcDNA3-TMPRSS2:ERG plasmids. Resistant clones were selected based on antibiotic resistance over 15–21 days in DMEM containing 300 μg/mL G418. Selected colonies were isolated, allowed to proliferate and characterized for ERG and TMPRSS2:ERG expression using RT-PCR and Western blot analysis.

The assays were conducted as described in reference (33) using purified ERG protein [expressed and purified using the T7-Impact System (New England Biolabs) as described previously (34)] and ERG transfected-HeLa nuclear extracts. An oligonucleotide (Eurogentec) probe nt −123 to −101 (5′-CCAGAGGAGGAAGTGTAGGAGCAGGT) was prepared by end-labeling double-stranded oligonucleotides with [³²P]ATP (2500 Ci/mmol) using T4 polynucleotide kinase (Promega) followed by G-50 column purification (GE-Healthcare). Radiolabeled oligonucleotide probes (100,000 cpm) were incubated with protein or nuclear extracts in a total volume of 20 μL for 30 minutes at 20°C. Binding buffer consisted of 100 ng/mL of poly(dI-dC)-poly(dI-dC), 20 mmol/L Tris-HCl (pH 7.9), 50 mmol/L KCl, 1 mmol/L MgCl2, 0.5 mmol/L EDTA, 0.5 mmol/L dithiothreitol, 0.02% NP-40 and 10% glycerol. DNA-protein complexes were resolved by electrophoresis for 1 hour at 200 V on a 5% nondenaturing polyacrylamide gel. The gels were dried, and signals were detected by exposure to autoradiography film. Nuclear extract was used as the negative control. In competitive binding assays, unlabeled oligonucleotides were added in 100 M excess. Oligonucleotides with mutated EBS were designed as follows: mutant 1 5′-CCAGAGGTAAAAGTGTAGGAGCAGGT, mutant 2 5′-CCAGAGGAGGAAGTGAAAAAGCAGGT and mutant 1+2 5′-CCAGAGGTAAAAGTGAAAAAGCAGGT. Supershift assays were conducted by adding 1 μL of polyclonal antibodies directed against ERG (sc-353 and sc-354, Santa Cruz Biotechnology).

ChIP assays were conducted as previously described (35) by using anti-ERG (sc-353, Santa Cruz Biotechnology). Briefly, formaldehyde cross-linked chromatin was sonicated and immunoprecipitated with either IgG (control) or the indicated Ab. ChIP analyses were conducted in at least triplicate using distinct DNA preparations. Immunoprecipitated DNA was purified and quantified by PCR. The targeted OPN promoter sequence (−269; −75) was amplified using the following primers: 5′-CATGGGATCCCTAAGTGCTC-3′, 5′-TGAGGTTTTCTGCCACTGCCC-3′. The target irrelevant sequence was amplified using primers: 5′-TGAGAGCAATGAGCATTCCGATG-3′ and 5′-CAGGGAGTTTCCATGAAGCCAC-3′.

Total RNA was extracted using the Nucleospin RNA II kit according to the manufacturer's protocol (Macherey-Nagel). One microgram of total RNA was reverse-transcribed using Superscript II RT (Invitrogen), random hexamers (Roche), and dNTPs at 42°C for 1 hour. OPN fragments were then amplified using the High Fidelity PCR Master kit (Roche) and the appropriate oligonucleotides (A: 5′-TGAGAGCAATGAGCATTCCGATG-3′, B: 5′-CAGGGAGTTTCCATGAAGCCAC-3′) using the following cycling parameters: initial denaturation at 94°C (5 minutes), then 30 cycles of 94°C (30 seconds), annealing at 60°C (30 seconds) and extension at 72°C (30 seconds), followed by a final extension at 72°C (7 minutes). The TMPRSS2:ERG fusion transcripts were amplified with the primers: TMPRSS2 forward primer 5′-CGCGAGCTAAGCAGGAGGC-3′ and ERG reverse primer 5′-GTAGGCACACTCAAACAACGACTGG-3′ described in (24). As an internal control, hypoxanthine–guanine phosphoribosyltransferase (HPRT) fragments were also amplified under the same conditions (forward primer: 5′-GCTGGTGAAAAGGACCTCT-3′ and reverse primer: 5′-AAGTAGATGGCCACAGGACT-3′). PCR products were analyzed by electrophoresis in a 2% agarose gel.

Cellular extract preparation and Western blot analysis were carried out as previously described (35, 36). Immunodetections were carried out using a polyclonal anti-ERG antibody (sc-353, Santa Cruz Biotechnology), a monoclonal anti-OPN antibody (sc-21742, Santa Cruz Biotechnology) and an anti-actin monoclonal antibody (Sigma).

Immunohistochemistry was carried out on an automated immunostainer (Benchmark XT) with the XT ultraview diaminobenzidine kit. The primary antibodies were directed against ERG (sc-353, Santa Cruz Biotechnology, dilution 1/50) or OPN (sc-21742, Santa Cruz Biotechnology, dilution 1/100). Antigen retrieval was conducted in Tris-EDTA buffer (pH 9) for 30 minutes at 95°C for ERG and in citrate buffer (pH 6) for 60 minutes at 95°C. Negative controls were realized by omitting primary antibodies. Counterstaining was done with hematoxylin and bluing reagent.

Statistical analysis was conducted using GraphPad Quick calcs (GraphPad Software). To analyze gene expression and the correlation between the expression of TMPRSS2:ERG and OPN, we used a 2 × 2 correlation table and Fisher's exact test, and Spearman's correlation coefficient. We considered results statistically significant at P < 0.05.

To investigate the possibility of ERG-mediated OPN transactivation, we undertook a comparative genomic bioinformatics analysis to identify regions conserved between species and to define EBS. Using the ECR browser, the 5′ flanking region of the human OPN start site was compared with mouse, dog, and rhesus macaque genomes (Supplementary Fig. S1). This analysis shows that the OPN proximal promoter is well conserved in all species analyzed, and 2 domains were defined: one from nt −289 to +6 bp and the other from nt −1100 to −920 kb. These 2 conserved domains were scanned using Transfac to find potential EBS composed of the 5′ consensus sequence -GGA/W-3′ (W:A or T). We found only 1 putative conserved EBS upstream of the OPN gene at nt −115 to −118 (EBS1). Another potential inverted EBS was identified at nt −108 to −111 (EBS2), but was found only in the mouse and rhesus macaque genomes. The computer search for transcription factor binding sites (Transfac analysis) also disclosed Runx2, SBF-1, C/EBP, and AP-1 sites, which have already been reported (30).

To test ERG transactivation, a series of sequential 5′-deletion mutant constructs were prepared from the region of the promoter sequence that confers efficient transcription of the OPN gene (37). The 5′-deletion sequences were cloned into the pGL3 reporter plasmid (Fig. 1A) and assayed for ERG responsiveness via luciferase activity in transient transfected HeLa cells (Fig. 1B). Moreover, since ETS-1 factors have been shown to regulate OPN promoter (30), we conducted luciferase assays using the same constructs (Supplementary Figs. S2A and B). As expected, ETS-1 regulate the OPN promoter. Since ERG protein overexpression has been associated with more than 50% of PCas, we conducted transfection assays in a biologically relevant cell type for studying OPN in PCa: PC3c cells that derived from the human prostate carcinoma PC3. Similar results were obtained compared with HeLa cell experiments (Fig. 1C). As shown in Figure 1B (and Supplementary Fig. S2A, assay using the plasmid pcDNA3-ETS-1), the deletion of nt −136 to −116 bp resulted in a dramatic loss of OPN promoter activity after ERG (or ETS-1) overexpression, whereas OPN expression was not affected after the deletions including nucleotides downstream from −136 bp. Luciferase activity increased in a dose–dependent manner with the plasmid pSG5-ERG (Fig. 1D) or with the plasmid pcDNA3-ETS-1 (Supplementary Fig. S2B). As expected, OPN expression was dramatically decreased after shRNA-ERG cotransfections compared with the shRNA control. These results showed that ERG transactivates the OPN promoter and this region, nt −136 to −116 bp, contains one or several essential cis-regulatory sites for ERG.

To investigate whether these EBS were direct ERG targets, we conducted an EMSA using a ³²P-labeled WT nt −126 to −101 oligonucleotide substrate and purified ERG protein (Fig. 2A). Banding patterns revealed ERG affinity for these binding sites (designated as S in Fig. 2A) compared with the probe alone. In the presence of ERG antibodies (sc-353 Ab and sc-354 Ab), there were fewer S complexes and specific antibody-supershift bands SS′ and SS" were formed. No supershift was observed using the IgG isotype. Similar banding patterns were obtained when EMSA assays were conducted in the presence of nuclear extracts from HeLa cells (Supplementary Fig. S3) and PC3c (Fig. 2B) transfected with pSG5 or with pSG5-ERG. A supershift SS′ was only detected with ERG-specific antibody in pSG5-ERG transfected cells.

In accordance with these results, the shift assay revealed ERG binding on the nt −126 to −101 sequence of the OPN promoter. ChIP assays were then conducted with transfected HeLa cells to confirm that the EBS identified by EMSA in this portion of the OPN promoter is able to recruit ERG in vivo (Fig. 2C). As expected, ERG proteins were tethered to the endogenous OPN promoter. No precipitation was detected with ERG antibodies in an irrelevant region (IR), validating the specificity of binding ERG to EBS. Together, these experiments clearly showed that ERG binds to the OPN promoter.

To further characterize which EBS are involved in ERG binding within the WT nt −126 to −101 oligonucleotide region, we used site-directed mutant probes and constructs (Fig. 3A) and analyzed them using EMSA (Fig. 3B and 3C) and luciferase-promoter assays (Fig. 3D). A gel-shift complex (S) was present with the wt probe itself and the mutant 2 probe, but not with the mutant 1 or mutant 1+2 probes (Fig. 3B). In contrast, this complex was absent in 100-fold excess of unlabeled wt and mutant 2 probes, but persisted with mutant 1 probe (Fig. 3C). Similar patterns were obtained with nuclear extracts of transfected HeLa cells (data not shown).

Next, to assess the functionality of the EBS, we investigated the effect of ERG overexpression using a luciferase reporter construct driven by mutated EBS. The OPN promoter fragment −136/+77, with ERG-responsive transcriptional activity similar to the intact 1518 bp promoter (Fig. 1B), was used as a template to introduce site-directed mutations in predicted EBS1 and EBS2 sequences, named OPN −136/+77 mutant1 and OPN −136/+77 mutant2, respectively, and transfected into HeLa cells (Fig. 3D). A mutation in nt −119 to −115 from GGAGGAAG to GGTAAAAG resulted in the suppression of ERG transactivation. In contrast, luciferase activity was not affected by mutations in nt −112 to −106 from GTAGGAG to GTCGGAG. Thus, these experiments suggest that ERG may play a direct role in OPN promoter activation through a very specific EBS located at nt −118 to −115.

A high-incidence recurrent fusion of TMPRSS2 with ERG has recently been highlighted in PCa. This fusion results in aberrant androgen-regulated ERG in prostate cells. We therefore examined endogenous OPN expression in a prostate cell line, PC3c, that we stably transfected with ERG and TMPRSS2:ERG (Fig. 4). Efficiencies of ERG and TMPRSS2:ERG overexpression were characterized in several subclones and compared with nontransfected cells and empty pcDNA3 vector-transfected cells by RT-PCR (Fig. 4A) and Western blotting (Fig. 4B). The molecular weight of ERG was expected to be 55 kD, whereas TMPRSS2:ERG was expected to be 49 kD, consistent with the deletion of 32 amino acids from the N-terminus (Fig. 4B). Results obtained with representative subclones were similar for several subclones (data not shown). Expression analysis showed that, compared with the nontransfected PC3c and empty pcDNA3 vector-transfected PC3c, OPN expression was greater in clones that overexpress ERG following transfection of ERG and TMPRSS2:ERG. Interestingly, ChIP assays revealed that ERG transcriptional factors were associated with the OPN promoter in clones of PC3c cells expressing ERG and TMPRSS2:ERG in which OPN was up-regulated but not in empty pcDNA3 clones (Fig. 4C). Together, these data suggest that ERG and TMPRSS2:ERG gene fusion products bind to the cis-regulatory domain of the endogenous OPN promoter in PC3c cells and transactivate this gene.

Samples of human primary PCas were assessed for OPN and TMPRSS2:ERG transcript expression using RT-PCR on total isolated RNA (Table 1): TMPRSS2:ERG fusions were detected in 71.4% (25/35) of the prostate tumors. This frequency of TMPRSS2:ERG fusions is in the range reported from studies conducted in Western countries (38, 39). Among these TMPRSS2:ERG+ tumors, 96% were also positive for OPN expression. Using a 2-tailed, cross-tabulated Fisher's exact test showed a significant association (P < 0.01) between TMPRSS2:ERG and OPN expression (Table 2).

Next, we examined the expression of ERG and OPN from human prostate tumors (Fig. 5A). Using immunohistochemistry, we observed the coexpression of ERG and OPN in areas of prostatic adenocarcinoma. Staining against ERG was nuclear with a moderate to strong signal in adenocarcinomatous glands as shown in Figure 5. OPN was detected in the cytoplasm of adenocarcinomatous glands. As a control, in normal prostatic glands, we did not observe ERG expression except in endothelial cells of vessels (Supplementary Fig. S4), whereas, as previously described (40), none or weakly stained OPN expression was observed in the prostatic stroma. Moreover, no increase of the OPN expression level was observed in higher Gleason Score samples (Supplementary Fig. S4 and data not shown; ref 40).

Among 16 samples analyzed immunohistochemically, there were 12 (75%) cases with OPN-positive cells and 9 (56%) cases with TMPRSS2:ERG fusion positive cells (Table 3). Spearman's correlation coefficient was used to evaluate the correlation between protein expression levels (referred as % of cells stained) of ERG and OPN in IHC (Fig. 5B). Our data showed that there was a significantly positive correlation between expression levels of ERG and OPN (r = 0.531; P < 0.05).

OPN is a SPP that has been found to facilitate anchorage of normal bone or cancer cells to mineralize tissue surfaces (41). This protein can be considered as one of the critical determinants in bone formation and tumor progression. The malignant and metastatic phenotype of human cancers has been correlated with elevated OPN expression, both at mRNA and protein levels (7). Uncovering transcriptional regulators of the OPN promoter has been stimulated by research seeking potential molecular targets that could modulate the metastatic phenotype and thereby constitute an effective therapeutic treatment. Several signaling pathways/transcription factors have already been proposed [for review, see ref. (42)]. In this study, we showed that ERG, a transcription factor associated with precartilaginous condensation and chondrogenesis preceding bone formation (43, 44), binds to the OPN promoter and transactivates this gene. Through EMSA experiments, a functional and conserved cis-regulated enhancer site was localized at nt −118 to −115. A mutation in this consensus binding site abolished ERG transactivation. In agreement with EMSA, ChIP assays revealed that ERG directly binds to the OPN promoter in cells. These results suggest that ERG is a critical transactivator of the OPN promoter. However, as noted (Supplementary Fig. S1), many transcription factors have DNA binding sites adjacent to the EBS sequence in the OPN promoter. Transcriptional activity of the ERG factor, as in other ETS members, is modulated by the cooperation with other key transcription factors such as AP-1, CBP-P300, Runx2, NFkB, Gata1, etc. (45–47). The combinatorial control of OPN regulation by the ERG factor and its partners is a fertile ground for ongoing and future investigations on ERG expression in the cell.

Recently, ERG has been discovered to be highly overexpressed in the majority of human prostate tumors, due to the fusion of TMPRSS2:ERG through chromosomal rearrangements. The consequences of the TMPRSS2:ERG gene fusion in prostate tumorigenesis initiation remains controversial (48) and studies on its prognostic value have produced contradictory results (28, 49–51). Furthermore, PCa is characterized by its high propensity—occurring in more than 80% of patients—to spread to skeletal tissue. However, the mechanisms underlying this preferential homing of prostate cells to bone tissues are only beginning to be understood. The high level of OPN expression is closely linked to the degree of malignancy and with the reduced survival of patients (40). OPN has been shown, along with other skeletal and matrix-associated proteins, to be expressed in cancer cells as major soluble factors stimulating the migration, survival, anchorage-independent growth and invasion of PCa cells (52, 53). In this context, we expected that the abnormal expression of ERG protein in prostate tumor cells via the TMPRSS2:ERG rearrangement would induce ERG overexpression and thereby affect the expression of ERG-target genes. Among these deregulated genes, we hypothesized that the OPN gene would increase through ERG overexpression. In this study, we show that, in PC3c cells stably transfected by ERG or TMPRSS2 (exon 1):ERG (exon 4; the most frequent fusion in PCa), OPN transcription is enhanced and that ERG binds to the OPN promoter in PC3c DNA. Proteins encoded by this TMPRSS2 (exon 1):ERG (exon 4) fusion transcript are truncated, lacking 32 amino acids from the N-terminus. Despite this truncation in the N-terminal region, whose functional role remains unknown (54), both ERG and TMPRSS2 (exon1):ERG (exon4) similarly regulate OPN expression. However, in normal prostate cells, endogenous ETS factors may regulate a set of genes. Therefore, in PCa cells, which exhibit the TMPRSS2:ERG fusion, the ERG proteins are strongly overexpressed causing a perturbation of transcriptional gene regulation. These changes could affect newly genes or ETS targets already regulated by endogenous ETS factors. For example, since ETS-1 is able to directly activate the OPN promoter, a competition could exist between ETS-1 and the overexpressed ERG factors for the OPN promoter binding and activation.

To test the biological relevance of our findings suggesting that TMPRSS2:ERG may increase OPN expression levels in prostate tumors, coexpression of OPN and ERG were assessed using RT-PCR and immunohistochemical analyses in specimens from patients diagnosed with PCa (mainly biopsies from prostatectomy with Gleason scores evaluated at 7). Despite the biological heterogeneity that characterizes PCa specimens, which can complicate molecular studies, a significant association was observed between TMPRSS2 (exon 1):ERG (exon 4) fusion and OPN expression. Moreover, tumor specimens showed similar immunostaining localization for ERG transcription and OPN. In this study, we only focused on the involvement of the most common fusion variant TMPRSS2 (exon1):ERG (exon 4) reported in OPN upregulation. However, this does not exclude the possibility that other fusion variants between androgen-dependent gene promoters and ETS family members are present in prostate tumor samples where TMPRSS2 (exon1):ERG (exon4) has not been detected.

OPN has multi-functional properties that promote cell survival, cell adhesion, and cell migration (42, 55). By binding to αvβ3 integrins, OPN protects cells from apoptosis through the activation and phosphorylation of the PI3-K/AKT pathway (56, 57). OPN activates various matrix-degrading proteases, such as matrix metalloproteinases (MMP) and the urokinase plasminogen activator (PLAU) that contribute to malignancies (58, 59). MMP3, MMP9, a disintegrin and metalloproteinase 19 (ADAM19) and PLAU were recently reported to be increased in benign prostate cell lines RWPE and PrEC (primary benign prostate epithelial cells) transfected with overexpressed ERG factors (60). They were identified as direct targets of ETS transcription factors. Based on our results, we can also suggest that the activation of the OPN by TMPRSS2:ERG mediates the matrix metalloproteinase pathway and thereby PCa progression.

In summary, the results presented here show that OPN is a target gene of ERG. The expression of the extracellular matrix protein OPN, which plays a crucial role in tissue remodeling, inflammation, tumor growth, angiogenesis and metastasis, is associated with TMPRSS2:ERG expression in prostate tumors cells and thereby participates in metastasis progression and aggressiveness. Small molecules that can prevent the interaction of TMPRSS2:ERG with the promoter of ERG-target genes in PCa cells may be an interesting avenue of research for developing metastasis prevention strategies.

No potential conflicts of interest were disclosed.

We thank Dr. Marc Aumercier for purified ERG protein; Drs. Yvan de Launoit and Marion Le Jeune for their useful suggestions and discussions, the local tumor tissue library/regional oncology reference center Tumorothèque, CRRC Lille (head: Prof. Copin) for providing frozen tissues.

This work was supported by grants from the CNRS, ANR-TecSan (Promocart 2006), La Ligue contre le Cancer (Comité du Pas-de-Calais) and INSERM PRO-A. S. Flajollet is supported by a fellowship from ANR-TecSan (Promocart 2006). T.V. Tian is recipient of Ph.D. fellowship from Institut Pasteur de Lille/Région Nord-Pas-de-Calais and Faculté de Médecine Henri Warembourg-Université du Droit et de la Santé Lille 2.

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