Molecular Validation of PACE4 as a Target in Prostate Cancer^,

Abstract

Prostate cancer remains the single most prevalent cancer in men. Standard therapies are still limited and include androgen ablation that initially causes tumor regression. However, tumor cells eventually relapse and develop into a hormone-refractory prostate cancer. One of the current challenges in this disease is to define new therapeutic targets, which have been virtually unchanged in the past 30 years. Recent studies have suggested that the family of enzymes known as the proprotein convertases (PCs) is involved in various types of cancers and their progression. The present study examined PC expression in prostate cancer and validates one PC, namely PACE4, as a target. The evidence includes the observed high expression of PACE4 in all different clinical stages of human prostate tumor tissues. Gene silencing studies targeting PACE4 in the DU145 prostate cancer cell line produced cells (cell line 4-2) with slower proliferation rates, reduced clonogenic activity, and inability to grow as xenografts in nude mice. Gene expression and proteomic profiling of the 4-2 cell line reveals an increased expression of known cancer-related genes (e.g., GJA1, CD44, IGFBP6) that are downregulated in prostate cancer. Similarly, cancer genes whose expression is decreased in the 4-2 cell line were upregulated in prostate cancer (e.g., MUC1, IL6). The direct role of PACE4 in prostate cancer is most likely through the upregulated processing of growth factors or through the aberrant processing of growth factors leading to sustained cancer progression, suggesting that PACE4 holds a central role in prostate cancer.

Introduction

Cancer cells are characterized by multiple genetic alterations that confer physiological changes, leading to uncontrolled division and ability to invade other tissues [1]. These acquired capabilities, namely cell proliferation, tissue invasion, and adhesion, are essential for malignant growth and progression. Numerous studies have associated the family of enzymes known as the proprotein convertases (PCs) to cancer [2–8]. PCs are serine proteases that cleave substrates at R-X-K/R-R↓ motif [9,10]. These processing events, resulting in the activation of protein precursors, occur at multiple levels of cell secretory pathways, and even at the cell surface. In mammalian cells, seven basic amino acid cleaving members of this family have been identified: furin, PACE4, PC1/3, PC2, PC4, PC5/6, and PC7. PCs show differential expression levels within tissues, ranging from ubiquitous (e.g., furin) to an endocrine-restricted expression (e.g., PC1/3 and PC2).

The first association of PCs with cancer was done by comparative studies of normal and cancerous cells showing higher expression of PCs in small cell lung cancer [11], non-small cell lung carcinoma [12], breast [13], colon [14], and head and neck [15] tumor cells or lower expression in ovarian cancer [16]. A correlation between expression of some PCs, namely furin and PACE4, and tumor cells' aggressiveness has been established for different cell types [8,15,17–20]. Thus, it is hypothesized that through processing events, cancer cells have a higher capacity (i) to remodel the extracellular matrix (e.g., MT1-MMP and MT2-MMP; [17,21,22]), (ii) to interact with their host microenvironment to favor adhesion (e.g., IL-1α, TNF-α, and E-selectin [7]); and (iii) to modulate their proliferation and differentiation (e.g., IGF-1 receptor [7], VEGF-C, -D [23,24], PDGF-A, -B [24,25], TGF-β [26]). One possible mechanism whereby PCs are critical in cancer progression could be on the basis of differential precursors processing by overexpressed PCs. Alternatively, PCs overexpression is required to sustain pathophysiological functions to maintain cancer cells immortality, for example to sustain the processing of required growth factors or proangiogenic factors. Together, these studies suggest that PCs processing events are directly linked to tumor development by allowing cancer cells to maintain some of their acquired capabilities.

Because the PC family of enzymes is very large and often responsible for redundant processing functions, more precise studies are required if we are to understand their specific roles in different cancer types. Thus, there is a need for potent, specific, and cell-effective inhibitors, either pharmacologic or molecular, for each member of this enzyme family. Pharmacological tools are limited as many lack the required specificity for single PCs [9,27–30]. For example, a study by Uchida et al. [31] showed that treatment of prostate cancer cell lines with the general PC inhibitor decanoyl-Arg-Val-Lys-Arg-chloromethylketone (dec-RVKR-CMK) inhibited prostate-derived factor processing and resulted in the loss of luminal cell phenotype and induction of basal cell phenotype as demonstrated by alternations in the expression of cytokeratins 8, 14, 18, and 19, markers of prostate epithelial cell differentiation. Because dec-RVKR-CMK inhibits all PCs, the study could not distinguish which PCs expressed in the prostate cancer cell lines were more important. To ask the question as to which PCs are the most important for cancer progression, molecular approaches targeting specific PC messenger RNA (mRNA) are closer to achieving this goal. There are various sequence-specific approaches taking advantage of either endogenous ribonucleases (e.g., small interfering RNA [siRNA] and phosphorothioate antisense oligonucleotide) or their intrinsic catalytic activity (e.g., deoxyribozyme and ribozyme) to drive the degradation of the targeted RNA [32].

In the present study we chose to examine the role of PCs in prostate cancer. In men, prostate cancer accounts for the highest incident cases and the second most deadliest cancer type [33]. Because disease diagnosis and treatment have been little changed in the past 30 years, it is possible that a better understanding of the molecular events leading to advanced prostate cancer states could lead to the development of new, efficient, and targeted therapies [34]. We first evaluated the implication of each PC on different types of prostate cancer biopsies and detected the unique PACE4 mRNA overexpression. Based on these findings, we created a DU145 cell line model, using the switch on/off adapter-hepatitis delta virus ribozyme (SOFA-HDVRz) as a RNA silencing tool, to investigate PACE4's contribution in androgen-independent prostate cancer cells. The results of this study, based on in vitro assays, gene expression, and proteomic profiling, as well as animal studies, led us to the conclusion that PACE4 plays a critical role in the progression of prostate cancer and therefore could be a novel therapeutic target for its treatment.

Materials and Methods

Patients and Samples

Prostate tumor samples were obtained from either St Louis and Bichat Hospital (Paris, France) or Tournan's clinic (Tournan en Brie, France), as previously described [35]. Each patient included has signed the sample informed consent form. Sample tissues from the 34 patients with clinically localized prostate tumors were obtained by removing clinically localized tumors by radical prostatectomy and classified according to their clinical stage using the TNM system [36,37]. After pathological examination, 18 samples were at stage pT2, whereas 16 samples were at stage pT3. The sample tissues from hormone-refractory recurrent prostate carcinoma were obtained from patients with metastatic disease at diagnosis. Because these patients were not amenable to radical surgery, they received endocrine therapy, either by classic androgen deprivation (orchidectomy or luteinizing hormone-releasing hormone agonist administration) or by maximal androgen blockade (castration combined with antiandrogen therapy). These patients relapsed, and their tumors became clinically androgen-independent. The 13 samples of these hormone-refractory recurrent prostate carcinomas were obtained during transurethral resection. Suspect areas were examined histopathologically in the surgery suite, and thick shave sections were taken for research purposes. These pre-selected tumor specimen sections were then sliced on each side in the laboratory and again subjected to pathological examination. Confirmed areas were carefully microdissected using a scalpel and the assistance of an experienced pathologist. Well-characterized matched normal prostate specimens (nine samples) from the 34 patients with clinically localized prostate tumors who underwent radical prostatectomy were used to assess basal target gene mRNA expression. Normal-looking areas of each surgical specimens were examined histologically for the absence of cancer cells and selected on their microscopic pathological criteria to avoid the inclusion of areas with benign hyperplasia. Normal specimens were then proceeded as previously described for tumor tissues.

Real-time Polymerase Chain Reaction on Tissues

The RNA was extracted from dissected tissues with an acid-phenol guanidium method, reverse-transcribed, and submitted to quantitative polymerase chain reaction (PCR) as previously described [35]. A genomic DNA and nontemplate control were included in each experiment. Samples and controls were tested in duplicate. The threshold cycle (C_t) numbers obtained from PCR amplifications were expressed as fold change in target gene expression (PACE4) relative to an endogenous control gene (PPIA; peptidyl prolyl isomerase A gene encoding cyclophilin A). The primer sequences for endogenous control gene PPIA (the peptidyl prolyl isomerase A gene encoding cyclophilin A) were described earlier [35]. The primer sequences were as follows: PACE4 sense, 5′-CAAGAGACCCAGGAGCATCCC-3′; PACE4 antisense, 5′-ACCCGCTGGTCCGAGTGCT-3′; furin sense, 5′-CCAGGATGAATCCCAGGTGCTC-3′; furin antisense, 5′-GGAGGGTGAAGAGTGCCGACC-3′; PC7 sense, 5′-CATCCAGGACATTGCACCCAAC-3′; and PC7 antisense, 5′-GGTTGCCATTCTCCACATCCG-3′.

In Situ Hydridization

The in situ hydridization protocol using digoxigenin-labeled complementary RNA probes have been described in details previously [38].

SOFA-HDVRz Design, Cloning, and In Vitro Cleavage Assays

Bioinformatic analysis. Human PACE4 complementary DNA (cDNA) sequence (M80482) was used for the determination of the optimal SOFA-HDVRz targeting site. The selection criteria were (i) the first nucleotide (nt) should be a G to form the required wobble bp within the P1 stem; (ii) after the 6 subsequent nt of the P1 stem, a 5 nt spacer was added; and (iii) a 12-nt biosensor stem was used to complete the design. Thus, the targeted site on cDNA was 5′ (GN₆)_P1-(N₅)_Spacer-(N₁₂)_Biosensor-3∲ (Figure 2A). We focused our search for optimal cleavage site within the first 1200 nt of the PACE4 cDNA and tested several potential targeting sites with the Ribosubstrates online software (http://www.riboclub.org/ribosubstrates) [39]. This integrated software searches in selected cDNA databases all potential substrates for a given SOFA-HDVRz. These potential substrates include mRNA with perfect matches with the catalytic RNA tested, but also the wobble bp and mismatches. Only SOFA-HDVRz with a human PACE4-specific biosensor and harboring a minimum of wobble bp were considered.

Figure 2.

Down-regulation of PACE4 expression in DU145 cells. (A) The secondary structure of both the “OFF” and “ON” conformation of the SOFA module (box) attached to the classic HDVRz. The biosensor (in red, BS) and the blocker (in green, BL) can be modified to be hybridized on a specific mRNA target (N indicates A, C, G, or U). After the recognition of the mRNA, the P1 stem (in purple, P1) issued from the sequence-specific hybridization of the HDVRz with the targeted RNA will allow the catalytic cleavage of the latter, as indicated by the arrow. (B) Autoradiogram of in vitro cleavage assays performed with either the SOFA-HDVRz or the chimeric RNA transcript tRNA^Val-SOFA-HDVRz both specific for PACE4 RNA. The control lane is the migration of the internally ³²P-radiolabeled 3800 nt PACE4 RNA alone. The BS, BL, and P1 nucleotide sequences of the PACE4-SOFA-HDVRz used are indicated in the Materials and Methods section. XC indicates the position of xylene cyanol after migration on the denaturing 5% PAGE gel. (C and D) Autoradiograms of Northern blot hybridizations showing levels of PACE4 (C), furin (D), and PC7 (D) mRNA in untransfected DU145 cells (control cell line) or in cell lines either transfected with the expression vector coding for the tRNA^Val-PACE4-SOFA-HDVRz alone (4-2) or cotransfected with a PACE4 cDNA expression vector (4-2 + PACE4) are shown. The bands corresponding to the 18S ribosomal RNA are shown for all blots. The size markers are shown on the right. The expression levels of PACE4 mRNA in transfected cell lines relative to the unstransfected DU145 cells were obtained by densitometric analyses using 18S ribosomal RNA as loading control (C). Values are mean ± SEM (n = 3). *P < .05.

Plasmid DNA constructs. As previously detailed [40], we used the expression vector ptRNA^Val/hygromycin containing the RNA polymerase III promoter tRNA^Val promoter for cellular applications. A PCR strategy was used to create a DNA template containing a 5′-KpnI restriction site and a 3′-blunt end. The sequences of the two complementary and overlapping DNA oligodeoxynucleotides used were as follows: sense, 5′-ATCCATCGGGTACCGGGCCAGCTAGTTT (GGCCTCTGCTAC)_BS(CAAC)_BLCAGGGTCCACC-3′ and, antisense, 5′-CCAGCTAGAAAGGGTCCCTTAGCCATCCGCGAACGGATGCCCA(ATCAAC)_P1ACCGCGAGGAGGTGGACCCTG (GTTG)_BL-3′. The underlined nt correspond to the KpnI restriction site, and those in parenthesis to the PACE4 specific biosensor (BS), blocker (BL) and P1 stem (P1) of the PACE4-SOFA-HDVRz. The purified and KpnI-digested PCR product was cloned in the expression vector previously digested with KpnI and EcoRV restriction enzymes. The vector used to restore PACE4 mRNA expression level contained the full-length PACE4 cDNA and a neomycin resistance gene (kindly provided by Dr J.W. Creemers [41]).

RNA synthesis. As previously described, radiolabeled PACE4 RNA was obtained from transcription of an XhoI-digested pcDNA3 vector containing a chimeric cDNA composed of the PC5/6A signal peptide linked to proPACE4 coding sequence [41] using T7 RNA polymerase with 50 µCi of [α-³²P]GTP [40]. The catalytic RNA were synthesized using a PCR-based strategy with the expression vectors to generate DNA templates containing a 5′-T7 RNA polymerase promoter using either the following sense primer 5′-TTAATACGACTCACTATAGGGACCGTTGGTTTCCGTAG-3′ or 5′-TTAATACGACTCACTATAGGGCCAGCTAGTTT-3′, complementary to the tRNA^Val promoter or the PACE4-SOFA-HDVRz, respectively (the underlined nucleotides correspond to the T7 RNA polymerase promoter sequence). The antisense oligodeoxynucleotide sequence used for both PCR was 5′-CCAGCTAGAAAGGGTCCCTTA-3′. After PCR, the purified products were used as templates for T7 RNA polymerase transcription of tRNA^Val-PACE4-SOFA-HDVRz or PACE4-SOFA-HDVRz, as described previously [40]. All products were purified on denaturing 5% (PACE4 transcript) or 7.5% (for PACE4-SOFA-HDVRz transcripts) PAGE.

Cleavage assay. The SOFA-HDVRz cleavage assays under single turnover conditions ([SOFA-HDVRz] > [PACE4 RNA]) were done at 37°C for 3 hours in a 10-µl reaction containing trace amount of radiolabeled PACE4 RNA and 1 µM of SOFA-HDVRz in reaction buffer containing 50 mM Tris-HCl, pH 7.5, and 10 mM MgCl₂. The reactions were stopped by the addition of loading buffer (97% formamide, 1 mM EDTA [pH 8.0], 0.025% xylene cyanol and 0.025% bromophenol blue), electrophoresed on denaturing 5% PAGE gel, and analyzed with a PhosphorImager (Amersham Biosciences, Sunnyvale, CA).

Cell Culture and DNA Transfection

Human cancer prostate cell lines DU145 were obtained from ATCC (Manassas, VA). Cells were maintained in Roswell Park Memorial Institute medium (RPMI 1640) supplemented with 5% fetal bovine serum(Wisent Bioproducts, St-Jean-Baptiste, Quebec, Canada). Cells were grown at 37°C in a water-saturated atmosphere in air/CO₂ (5%). Cells were transfected using Lipofectamine 2000 (Invitrogen, Burlington, Ontario, Canada) as per manufacturer's instruction and were selected for resistance to hygromycin B (Invitrogen) at 125 µg/ml, with 200 µg/ml of neomycin for double-transfected cells. The stable cell line transfected with the SOFA-HDVRz expression vector was named 4-2, whereas the 4-2 cell line stably transfected with the PACE4 expression vector was named 4-2 + PACE4.

Northern Blot Hybridizations

RNA extractions and Northern blots were performed as previously described [40]. The 1066-bp cDNA for human furin probe was obtained by digestion of the full-length clone with XhoI enzyme [42]. A 456-bp cDNA fragment of PACE4 was cloned in pGEM-T easy vector system (Promega, Madison, WI) by reverse transcription-PCR on DU145 total RNA with specific primers [13]. This vector was subsequently used for probe transcription. For the PC7 probe, we used a 285-bp rat cDNA [43], and for bovine 18S ribosomal RNA probe, a 600-bp cDNA [44] was used. The ImageJ software 1.37v (http://rsb.info.nih.gov/ij/) was used for all densitometric analyses.

Microarray Analysis

Sample preparation. RNA from DU145 or 4-2 cell pellets (in duplicate) was extracted using TRIzol reagent (Invitrogen) according to the manufacturer's instructions and purified using RNeasy mini kit (Qiagen, Mississauga, Ontario, Canada). The quality of the RNA was assessed using a 2100 Bioanalyzer with the RNA 6000 Nano LabChip kit (Agilent Technologies, Mississauga, Ontario, Canada).

Data processing. The microarray hybridization and the data collection were performed on the Functional Genomics Platform of the McGill University and Genome Québec Innovation Centre (Montréal, Quebec, Canada) using Illumina whole genome Human WG-6 v3.0 bead chip (38,275 genes). The processing of data was done with the software FlexArray v1.2 (http://genomequebec.mcgill.ca/FlexArray/). Briefly, the raw data were preprocessed and normalized for each array using the “lumi” algorithm. Then, the mean fold changes of normalized data were calculated for DU145 versus 4-2. Statistical analysis was performed using the Significance Analysis of Microarrays method, which uses a modified t test statistic and sample-label permutations to evaluate statistical significance [45].

Real-time PCR validation. The real-time PCR assays were carried out on the Mx3000P quantitative PCR system using the Brilliant II SYBR Green kits (Stratagene, Cedar Creek, TX). Briefly, the RNA was DNase I-treated (Invitrogen), reverse-transcribed using Superscript II reverse transcriptase (Invitrogen), and RNaseH-treated (Ambion, Austin, TX) before quantitative PCR. The reaction mixtures combined 1 µg of cDNA with 12.5 µl of 2x master mix containing primers at a 300-nM final concentration. The PCR profile was one cycle for 10 minutes at 95°C, 40 cycles of 30 seconds at 95°C, 1 minute at 60°C, and 30 seconds at 72°C. Primers were designed using Beacon Designer software (http://www.premierbiosoft.com). The sequences of primers used were as follows: SERPINE2 (NM_006216), sense 5′-CCCACTTCAATCCTCTGTCTCTC-3′ and antisense 5′-CCCAGGACCGACGCAATC-3′; CDK6 (NM_001259), sense 5′-GCCGCCCAGCCAGAACAC-3′ and antisense 5′-CTCCAGATAGCAATCCTCCACAGC-3′; IGFBP6 (NM_002178), sense 5′-GCGTGGAGGAGGAGGATGG-3′ and antisense 5′-CTTGGGCGGATGGCACTG-3′; TGFBI (NM_000358), sense 5′-GTGCGGCTGCTGGCTCTC-3′ and antisense 5′-GCTGGTAGGGCGACTTGGC-3′; CCNA1 (NM_003914), sense 5′-YGAAGAAGCAGCCAGACATC-3′ and antisense 5′-AAGTTGACAGCCAGATACAGG-3′; IGFBP3 (NM_001013398), sense 5′-TTCTGCTGGTGTGTGGATAAG-3′ and antisense 5′-GGCGTCTACTTGCTCTGC-3′; THBS1 (NM_003246), sense 5′-GGACTCGCTGTAGGTTATGATG-3′ and antisense 5′-CGGCTGCTGGACTGGTAG-3′; MUC1 (NM_002456), sense 5′-CTGGTGCTGGTCTGTGTTC-3′ and antisense 5′-GCTGCCCGTAGTTCTTTCG-3′; and ACTB (NM_007393), sense 5′-GGGAAATCGTGCGTGACATCAAAG-3′ and antisense 5′-CATACCCAAGAAGGAAGGCTGGAA-3′. For all PCRs, standard curves, dissociation curves, and migration of PCR products on acrylamide gels were done to confirm the specificity of the products. Moreover, two PCRs were performed as negative controls: one without cDNA template, and the other using reverse transcription reaction that never received SuperScript II reverse transcriptase enzyme. Results were estimated as C_t values, and the fold change of the expression of target gene in DU145 compared with 4-2 cells was expressed as 2^ΔΔC_t. The ΔC_t was obtained from the subtraction of the mean C_t for the target with the mean C_t for the internal control gene (β-actin). The Δ9C_t is the mean difference between the ΔC_t obtained from DU145 RNA extracts minus ΔC_t obtained from 4-2 RNA extracts. All PCRs were repeated twice for each set of primers.

Conditioned Growth Medium Preparation

Four 10-cm plates were seeded with 1.0 x 10⁶ cells (DU145 or 4-2 cells) in complete growth media. The next day, the cells were washed twice with PBS, and the growth medium was replaced with 5 ml of serum-free RPMI for 1 hour. The conditioned media were collected, filtered through 0.45-µm Acrodisc syringe filter units (Pall Corporation, Ville St. Laurent, Quebec, Canada), and kept on ice. The filtrates were concentrated to a final volume of 350 µl with Amicon Ultra centrifugal filter devices with a 3-kDa molecular weight cutoff (Millipore, Bedford, MA). Fresh RPMI was treated the same way as a control for cell experiments.

MTT Proliferation Assay

The cell proliferation was measured by the colorimetric MTT assay (thiazolyl blue tetrazolium bromide; Sigma-Aldrich, Oakville, Ontario, Canada). Briefly, cells were seeded in 96-well plates (BD Biosciences, Mississauga, Ontario, Canada) in triplicate with 100 µl of a 5.0 x 10⁴-cells/ml cell suspension in complete growth medium. The following day, cells were carefully washed twice with PBS twice and were incubated with 50 µl of either fresh RPMI, concentrated RPMI, DU145 or 4-2 concentrated conditioned growth media (DU145 and 4-2 CM, respectively). After 48 hours, 12.5 µl of an MTT solution (5 mg/ml in PBS) was added to each well for 4.5 hours at 37°C/5% CO₂. The medium was then carefully discarded, and the cells were solubilized with 100 µl of isopropanol/0.04N HCl solution. The absorbance was measured at a wavelength of 550 nm with a reference at 650 nm in microplate reader (SpectraMax190; Molecular Devices, Sunnyvale, CA).

Cell Count Assay

Cells were plated at a density of 5.0 x 10⁴/well in six-well plates (BD Biosciences) in duplicates. Complete growth medium was changed after 48 hours. After incubation, cells were washed in PBS, trypsinized, and counted after staining in 0.4% (wt/vol) trypan blue solution (Sigma-Aldrich). Only viable cells were counted in duplicate.

Colony Formation Assay

Cell lines were seeded in six-well plates (BD Biosciences) at a density of 300 cells/well in triplicate. Cells were maintained in complete growth medium for 10 days. After colony formation, medium was discarded, and cells were washed once with PBS. Colonies were fixed and stained in 5 mg/ml methylene blue/50% methanol solution for 10 minutes. Excess of staining solution was removed carefully with distilled water and the plates were dried overnight before scanning with Li-Cor Odyssey Infrared Imaging System (Li-Cor Biosciences, Lincoln, NE). Scanned images were analyzed with ImageJ software 1.37v (http://rsb.info.nih.gov/ij/) to measure the total particle area.

Flow Cytometric Analysis of Apoptosis without Cell Fixation

Propidium iodide DNA staining. Cell lines were seeded in 10-cm plates (5.0 x 10⁵ cells) in complete growth media and grown overnight. Afterwards, cells were washed three times with PBS, incubated in fresh complete growth medium for 12 hours, and harvested with trypsin (including the supernatant). The cell pellets were washed once with PBS, centrifuged, and resuspended in 50 µl of PBS. A chromatin decondensation treatment was performed using 450 µl of H-buffer/Triton X-100 (20 mM HEPES, pH 7.2, 0.16 M NaCl, 1 mM EGTA, and 0.05% Triton X-100) warmed at 37°C for 1 minute on ice. After this incubation, cells were gently centrifuged for 5 minutes and 450 µl of the supernatant was discarded. Cell pellets were then gently resuspended with the remaining supernatant, and the DNA staining was performed using 450 µl of H buffer containing 10 µg/ml of RNAseA (Sigma-Aldrich) and 10 µg/ml of propidium iodide (PI) (Sigma-Aldrich) for 30 minutes at room temperature in the dark.

Fluorescence-activated cell sorter analysis. After the PI staining, cells were analyzed on a FACScan cytometer (BD Biosciences). Forward and side scatter signals were used to establish the live gate to exclude debris and cell clumps. A minimum of 10,000 gated events by sample were acquired. The fluorescence of PI was collected in red channel on a linear scale (FL3). A second live gate was set using the FL3-A and FL3-W parameters of the doublet discrimination module, allowing single cell measurements. Fluorescence intensity distribution was analyzed with the CellQuest software (BD Biosciences), whereas the distribution of nonapoptotic cells into the phases of the cell cycle was determined using the ModFit software (Verify Software House, Topsham, ME).

Human Tumor Xenograft Models

Four-week-old female athymic nude mice (nu/nu; Charles River Laboratories, Wilmington, MA) were inoculated subcutaneously at the opposite sides of the flank with 3.0 x 10⁶ cells per inoculum (five mice/group). Cells were grown in complete media and harvested at their exponential growing state. Mice were housed under pathogen-free conditions, and the implantations were done under anesthesia conditions in laminar flow hood. Xenografts were measured three times per week, and volume (V) was determined by this equation: V = (L x W²) x 0.5, where L is the length and W is the width of the xenograft.

Immunohistochemistry

Subcutaneous tumors derived from DU145 and 4-2 cell lines were eradicated, formalin-fixed, and paraffin-embedded. Immunohistochemical staining was carried out on 3-µm sections by the Ventana NexES autostainer with the enzyme-labeled biotin-streptavidin system and the solvent-resistant DAB Map detection kit (Ventana Medical System, Tucson, AZ) using either p53 (1:200), MUC1 (epithelial membrane antigen; 1:200) or CDK4 (cyclin-dependent kinase 4; 1:100) anti-human monoclonal antibodies or prostate-specific antigen (PSA; 1:10,000) anti-human polyclonal antibody (Dako Canada, Burlington, Ontario, Canada; Thermo Fisher Scientific, Fremont, CA). The slides were counterstained with hematoxylin. The images were acquired using an Axioskop 2 phase-contrast microscope (Carl Zeiss, Thornwood, NY) and processed using Image Pro software (Media Cybernetics, Bethesda, MD).

Mass Spectrometry Analysis of Conditioned Media

Sample preparation. The DU145 and 4-2 conditioned growth media were prepared as described earlier in this section, except that the concentrated media were washed in centrifugal filter device with 12 ml of water before being collected and lyophilized. The samples were then solubilized with 100 µl of acidified water-0.1% trifluoroacetic acid. Acetone precipitation was performed with a ratio 4:1 (vol/vol). After a centrifugation at 13,000 rpm at 4°C for 30 minutes, pellets were solubilized in 50 mM pH 8.0 ammonium bicarbonate (NH₄HCO₃), a reduction of cysteine residues with a 45-mM dithiothreitol solution was performed for 15 minutes at 50°C, and a subsequent alkylation was performed with a 100-mM iodoacetamide solution for 15 minutes at room temperature in the dark.

Trypsin digestion. The samples were placed on ice for 30 minutes in 50 µl of proteases solution (sequence grade-modified trypsin [Promega] at 0.02 mg/ml in 25 mM NH₄HCO₃). Digestions were performed overnight at 37°C. The samples were then desalted and concentrated using C18 ZipTips (Millipore) before mass spectrometry analyses.

Matrix-assisted laser desorption/ionization tandem time-of-flight (MALDI-TOF/TOF) analyses. The MALDI TOF/TOF mass spectrometer (Ultraflex II; Bruker Daltonik, Bremen, Germany) used in this study was equipped with an Nd:YAG laser (355 nm) with a repetition rate of 200 Hz. Mass spectra were the sums of 8000 shots acquired in positive ion reflectron TOF mode. The acceleration potential was 25 kV. For MS² analysis, 2000 shots were acquired for parent ion selection and 10,000 shots for fragments selection. For MALDI analysis, the matrix used was α-cyano-4-hydroxycinnamic acid ACN: H₂O/trifluoroacetic acid 0.1% (7:3). Calibration was performed with peptides calibration standard (Bruker Daltonik). Samples underwent software processing with FlexAnalysis 2.4 and Biotools 2.2 (Bruker Daltonik) for baseline subtraction and peaks detection.

Protein identification. Protein identification was performed under the MASCOT sequence query search program using the SwissProt database filtered for the taxonomy Homo sapiens. A tolerance of 1 Da for peptide and 0.6 Da for tandem mass spectrometry (MS/MS) was set. Only protein sequences with a MOWSE score higher than 36 (indicating significant homology or identity) and identified in several samples representing at least two significant MS/MS were considered. Methionine oxidation and acetylation of N-terminals were defined as variable modification.

Statistical Analyses

All experiments were repeated at least three times, except for human tumor xenografts and microarray analysis. Comparisons between the test group and controls of this study were done using Student's t test.

Results

PACE4 Expression in Clinically Localized Prostate Tumors

We tested the expression of PCs in 47 primary prostate tumor samples obtained from patients undergoing surgery. The majority of the patients (i.e., n = 34) had clinically localized tumors and 13 had hormone-refractory recurrent prostate carcinoma. All collected samples were subjected to histopathologic examination. Only tissues where all epithelial cells were neoplastic were dissected and used for further studies. Real-time PCR was used to evaluate PACE4, furin, and PC7 mRNA expression levels in prostate tumor tissues using nine matched normal prostate tissues as a reference (Figure 1A). The tissues were grouped into similar clinical stages based on TNM system as 18 pT2 samples (tumors strictly confined to the organ), 16 pT3 samples (tumors with extracapsular extension), and 13 hormone-refractory samples (tumors no longer responsive to endocrine therapy). The mean relative PACE4 mRNA expression levels (Figure 1A) were significantly higher for all cancer groups (pT2, 4.0 ± 0.1; pT3, 4.3 ± 0.2; and hormone refractory, 2.8 ± 0.1), when compared with the mean level found in normal prostate tissues (1.8 ± 0.1). Real-time PCR for the other PCs showed that furin, PACE4, and PC7 were the most expressed PCs in normal prostate tissues. However, only PACE4 mRNA expression increased in tumor tissues, whereas the others showed a generalized reduction in cancer tissues. This higher PACE4 expression, particularly in epithelial cells, was directly assessed by an in situ hydridization. Normal prostate tissues showed the expected epithelial cell distribution of PACE4 mRNA. However, tumors showed disorganization of the tissue structure, with a higher PACE4 expression and even cells invading the stroma (Figure 1B).

Figure 1.

Relative mRNA expression levels of PCs in normal and tumoral prostate tissues. (A) Real-time PCRs from tissue specimens were performed with PACE4, furin, PC7, and PPIA specific primers. Nine normal prostate tissues and 47 prostate tumors (separated into three groups) were used for the comparative analyses. Clinical stage classification of prostate tumors, based on TNM system, was as follows: pT2, tumors strictly confined to the organ (18 samples); and pT3, tumors with extracapsular extensions (16 samples). Hormone-refractory prostate tumor tissues (13 samples) were issued from patients who relapsed after receiving endocrine therapy. Values are mean ± SEM. *P < .05, **P < .001. (B) PACE4 in situ hybridization in epithelial cells from normal and tumoral prostate tissues (arrows indicate purple staining corresponding to PACE4 mRNA).

Because PACE4 was the only member of the PC family to be overexpressed in prostate tissues, we investigated its expression levels in the Oncomine database (http://www.oncomine.org). As summarized in the Table 1, six independent studies reported a significant overexpression of PACE4 (P < .001) in prostate cancer tissues compared with corresponding normal tissues. Moreover, in five of the six studies, the PACE4 overexpression rank was within the top 2%, whereas the remaining was in the top 7%. It is noteworthy to report that furin, PC5/6, and PC7 did not show an up-regulation of their expression within those studies. These microarray studies, combined with our real-time PCR results, strongly suggest that PACE4 is associated with prostate cancer (i.e., potential biomarker or therapeutic target). Although these studies only measured PCs mRNA expression levels in prostate cancer tissues and not their enzymatic activity, we still decided to focus on PACE4 for this study.

Table 1.

PACE4 Is Upregulated in Prostate Cancer according to Oncomine Database^*.

Study	Description	Fold Change	P	Number of Measured Genes	Overexpression Gene Rank	Reference
1	Prostate carcinoma (n = 8) vs prostate gland (n = 4)	1.951	1.58e - 4	4564	7 (in top 1%)	Cancer Res. 2001;61(15):5692–5696
2	Prostate carcinoma (n = 61) vs prostate gland (n = 39)	2.190	9.12e - 15	10,021	35 (in top 1%)	Proc Natl Acad Sci USA. 2004;101(3):811–816
3	Prostate carcinoma (n = 65) vs prostate gland (n = 23)	1.246	4.65e - 7	7820	89 (in top 2%)	J Clin Oncol. 2004;22(14):2790–2799
4	Prostate carcinoma (n = 44) vs prostate gland (n = 13)	1.271	7.56e - 5	12,427	113 (in top 1%)	Cancer Res. 2006;66(8):4011–4019
5	Prostate carcinoma (n = 7) vs prostate gland (n = 6)	2.627	5.93e - 4	19,079	250 (in top 2%)	Cancer Cell. 2005;8(5):393–406
6	Prostate adenocarcinoma (n = 27) vs prostate gland (n = 8)	1.349	5.30e - 4	17,358	1202 (in top 7%)	Cancer Res. 2003;63(14):3877–3882

^*Available at: http://www.oncomine.org.

Rational Design of SOFA-HDVRz Targeting PACE4 mRNA

SOFA-HDVRz as a gene silencing tool. Considering the high expression level of PACE4 mRNA in tumoral prostate tissues, we decided to evaluate the contribution of this enzyme to the maintenance of the tumorigenic state in vitro and in vivo. Because there are no PACE4-specific and potent inhibitors available [9,46], we opted for a highly specific molecular approach targeting PACE4 mRNA. We previously showed that hepatitis delta virus ribozyme (HDVRz) could be a powerful and specific silencing tool in cultured cells [40]. Here, we tested a new generation of HDVRz harboring a switch on/off adapter (SOFA-HDVRz), resulting in an enhanced in vitro specificity and potency [47,48]. As shown in Figure 2A, the SOFA-HDVRz has three new features over the classic HDVRz. First, a blocker stem of 4 nt prevents the formation of the P1 stem, essential for the cleavage event. A biosensor of 12 nt at the 5′ end of the blocker stem compels the SOFA-HDVRz to maintain an inactive state (“OFF” conformation) until the right complementary RNA hybridizes. This sequence-specific hybridization releases the P1 stem from the blocker, allowing the formation of the P1 stem.

SOFA-HDVRz design. We used the complete sequence of human PACE4 cDNA (GenBank database accession number M80482) to design a selective SOFA-HDVRz against PACE4 mRNA. The bioinformatics analysis of the selected sequence reveals no other matches in the entire genome than that of PACE4. The exquisite specificity of the SOFA ribozyme approach has been validated [49]. Figure 2B shows the in vitro cleavage of PACE RNA with a selected PACE4-SOFA-HDVRz cleaving into the prodomain coding sequence of the transcript at position 519. As shown on the autoradiogram of the PAGE gel, the reaction generated only one cleavage product that could be seen after the addition of the PACE4-SOFA-HDVRz. To confirm the specificity of this molecule, we firstly tested the same ribozyme against a radiolabeled human furin transcript, and secondly, we used a catalytically inactive mutant (C76A) of the SOFA-PACE4-HDVRz against the PACE4 transcript [47]. In both cases, no cleavage product could be observed, confirming the capacity of the PACE4-SOFA-HDVRz to selectively cleave its target in vitro.

Down-regulation of PACE4 mRNA in DU145 Cells

Expression vector. We used an expression vector containing the tRNA^Val promoter to produce the PACE4-SOFA-HDVRz into transfected cells [40]. To insure that the tRNA^Val part did not interfere with the catalytic activity of the associated SOFA-HDVRz, we performed an in vitro cleavage assay by incubating the chimeric transcript (tRNA^Val-SOFA-HDVRz) with an internally radiolabeled PACE4 RNA (Figure 2B). As seen on the autoradiogram of the PAGE gel, the addition of the tRNA^Val motif did not modify neither the specificity nor the level of the cleavage activity of the PACE4-SOFA-HDVRz.

Northern blots. We transfected the PACE4-SOFA-HDVRz expression vector (named ptRNA^Val-PACE4-SOFA-HDVRz) into DU145, a highly invasive, androgen-independent prostate epithelial tumor cell line. These cells were chosen for their high level of PACE4 expression, aggressiveness, and ease of transfection when compared with the androgen-dependent LNCaP or the androgen-independent PC-3 cells. Stable cell lines transfected were established by the selection of clones resistant to hygromycin B. Northern blot analyses on total RNA extracts were performed for wild-type DU145 (DU145), DU145 transfected with ptRNA^Val-PACE4-SOFA-HDVRz (cell line 4-2), and 4-2 + PACE4 cells, which are 4-2 cell lines cotransfected with the PACE4 cDNA expression vector. As seen in Figure 2C, the PACE4 mRNA expression level in the SOFA-HDVRz transfected cell line was significantly reduced when compared with the untransfected cells. These levels were partially reestablished by the overexpression of PACE4 cDNA. The graph shows a densitometric analysis using 18S ribosomal RNA as loading control performed to quantify the mRNA expression levels in these clonal cell lines using wild type DU145 cells as reference (0.3 ± 0.1 and 0.75 ± 0.06 for 4-2 and 4-2 + PACE4, respectively). The mRNA expression levels of two other endogenous PCs confirmed the specificity of the PACE4-SOFA-HDVRz cleavage. Levels of furin and PC7 mRNA (Figure 2D) remained unchanged in the 4-2 cells, confirming the reduction of PACE4 expression without affecting the expression of other endogenous PCs. Because the autoradiograms suggested a slight variation in the expression of these two mRNA in the 4-2 + PACE4 cell line, we further examined these cell lines by quantitative PCR, confirming that no significant differences are observed.

Gene Expression Profiling in DU145 Cells

With the aim of better understanding the general contribution of PACE4 on prostate cancer growth, we performed a comparative gene expression analysis by DNA microarray screening with DU145 and 4-2 cells (Figure 3A). The gene expression levels measured by microarrays were validated by real-time PCR for eight genes covering a broad range of fold changes (Figure 3B). For all the genes tested, the quantitative PCR results (black bars) were either similar or higher than microarray data, confirming that microarray results (white bars) were reliable or even underestimated, at least for those tested. Of the 38,275 genes tested, 59 genes (0.15%) showed a two-fold decrease, whereas 62 genes (0.16%) had a two-fold increase in 4-2 cells (P < .05). These genes correspond to the dots found in the higher left and right quadrants of the volcano plot, respectively (Figure 3A; Tables W1 and W2).

Figure 3.

Microarray analysis and real-time PCR. (A) Volcano plot of differentially expressed genes in DU145 versus 4-2 cells measured with Illumina human expression bead chips. Vertical dashed lines show thresholds values for the fold change, whereas the horizontal dashed line indicates the threshold value for the P value (.05). Only genes with P values higher than the threshold with a log₂(fold change) lower than -1 or higher than 1 were include in Tables W1 and W2, respectively. (B) Confirmation of values obtained with microarrays by real-time PCR. Amplification and detection of cDNA were performed using SYBR Green I dye. The expression levels, measured by real-time PCR (black bars), were normalized using β-actin as internal control. The white bars indicate values obtained from microarray analyses. (C) Classification of differentially expressed genes in 4-2 cells compared with DU145 according to their biologic processes, cellular localizations, and molecular functions. The PANTHER classification system (http://www.pantherdb.org) had no attributed gene ontology grouping for nine downregulated genes and six upregulated genes.

The PANTHER (Protein Analysis THrough Evolutionary Relationships; http://www.pantherdb.org) system was used to group the gene listed in Tables W1 and W2 according to their biologic processes, cellular localizations, and molecular functions using the gene ontology classification (Figure 3C). Among the 59 downregulated and 62 upregulated genes, some could not be associated with any group into the database (9 and 6, respectively), whereas the others were distributed among multiple categories in comparable proportions between the two set of genes. Genes associated to cell communication, cellular processes (including cell cycle, cell growth, cell motion, and cell proliferation-associated genes), developmental, and metabolic processes are part of the major biologic functions affected by PACE4 down-regulation. The analysis of the cellular localization suggests a central role of PACE4 into the processing of various intracellular, membrane-bound, protein complexes and extracellular proteins. Finally, the major molecular functions affected by PACE4 down-regulation in 4-2 cells include cellular binding, transcription regulator, and catalytic (enzymatic) activity.

Although many of the listed genes have known biologic functions, not all proteins encoded by these genes are known or potential PACE4 substrates. Some interesting possibilities were observed, such as the analysis of CDH1 amino acid sequence (i.e., a cadherin) revealed that it is a potential PACE4 substrate because it necessitates a proteolytic processing event at the ₁₄₉LRRQKR↓DW₁₅₆ site of its propeptide domain to become biologically active [50]. Moreover, PTHLH also carries a potential PACE4 processing site at the ₃₁SRRLKR↓AV₃₈ of its propeptide region [51]. However, it is not excluded that downregulated genes (e.g., MUC1, a known prostate cancer oncoprotein [52]) might be indirectly affected by PACE4 expression levels without being a direct substrate harboring a PC processing [53].

PACE4 Is Important for the Activity of Growth Factors

Considering the previous links established between PCs expression, growth factors, and tumor progression, we measured cell proliferation activity by generating concentrated conditioned media issued from either untransfected DU145 or the 4-2 cell line. These growth media were incubated with DU145 cells for 48 hours, and proliferation state was measured by adding the MTT compound. As shown in Figure 4A, concentrated conditioned growth medium issued from DU145 cells (DU145 CM) stimulated the proliferation of DU145 cells in a similar range than the complete growth medium RPMI + FBS did (OD_550nm - OD_650nm = 0.44 ± 0.02 and 0.40 ± 0.02, respectively). This higher proliferation was not measured in cells incubated with conditioned growth medium issued from the 4-2 cell line (4-2 CM), where the OD_550nm - OD_650nm value (0.31 ± 0.02) was similar to that obtained with cells incubated with either RPMI or concentrated RPMI treated in the same way than the concentrated conditioned media, i.e., 0.28 ± 0.02 and 0.34 ± 0.03, respectively. Thus, these results suggest that PACE4 is important for the activity of growth factors that maintain a high proliferative status.

Figure 4.

Proliferation of DU145 cells with reduced PACE4 expression. (A) MTT proliferation assay with conditioned media. DU145 cells (5.0 x 10³/well in triplicate) were incubated either with RPMI, concentrated RPMI (RPMI conc), DU145 conditioned medium (DU145 CM), or 4-2 CM for 48 hours. After treatment, MTT solution was added for 4.5 hours, and OD_550nm - OD_650nm of solubilized cells was measured and compared using untreated cells (RPMI) as reference. Values are mean ± SEM (n = 7). (B) Time-dependent proliferation of untransfected DU145 cells (DU145) or expressing tRNA^Val-PACE4-SOFA-HDVRz with or without partially reestablished PACE4 expression (4-2 and 4-2 + PACE4, respectively) was assessed by counting total cell number after 24, 48, 72, and 96 hours of incubation time. Values are mean ± SEM (n = 3). (C) Colony formation after 10 days of incubation of DU145, 4-2, and 4-2 + PACE4 cell lines. Culture plates with fixed and stained cells were scanned, and the total particle area was measured. Typical result is illustrated for each cell line above in the upper part of the panel. The relative total area of the transfected cell lines compared with the control cell line is shown. Values are mean ± SEM (n = 9 for DU145 and 4-2 + PACE4 and n = 7 for 4-2). *P < .05.

PACE4 Down-regulation Slows DU145 Proliferation In Vitro

To further characterize the cell proliferation reduction observed with conditioned media experiments, we counted the total cell number of our stable cell lines at different times. As seen in Figure 4B, the results showed a significant reduction of proliferation for the 4-2 cell line (≈200,000 ± 10,000 cells) when compared with untransfected DU145 (≈380,000 ± 40,000 cells) 96 hours after the initial plating. It should be noted that these cell count experiments were carried out in the presence of complete growth media, which includes FBS; otherwise, the 4-2 cell line is very difficult to maintain and cell counts could not be obtained at the 72- and 96-hour time points. The observed reduced cell counts were partially reversed in the cell line 4-2 + PACE4 (≈280,000 ± 30,000 cells). We also performed an in vitro clonogenic assay on the same cell lines to detect the proportion of cells that retained the capacity to form colonies (Figure 4C). The results of this assay correlate the proliferation of DU145 with PACE4 expression levels, as we observed a 68% reduction of the colony growth in 4-2 cells when compared with wild-type DU145, whereas this colony formation capacity was partially restored in 4-2 + PACE4 cells (16% reduction when compared with the same cells).

PACE4 Inhibition Prevents Tumor Growth in Xenograft Tumor Model

We tested the ability of our experimental cell lines to grow as tumors in the nude mouse model. We injected 3 x 10⁶ cells subcutaneously in nude mice flanks and measured tumor volumes 1 week after injection. As shown in Figure 5A, the reduction of PACE4 mRNA expression level reduced dramatically the ability of the 4-2 cell line to induce tumor growth, whereas untransfected DU145 cells were able to develop into well-defined tumor masses. After 31 days, subcutaneous tumors were removed from nude mice for immunohistochemical staining of p53, MUC1, CDK4, and PSA proteins. We used p53 as a marker of tumoral cells because DU145 cells strongly express a double-mutant inactive form of this protein [54]. As shown in Figure 5 (B and C), the tumor derived from the 4-2 cell line contained a much lower number of tumoral cells when compared with the tumor derived from the DU145 cell line. We confirmed that the PSA protein was not expressed [55], and it was therefore included as a negative control (Figure 5, H and I). As suggested by the comparative gene expression analysis of DU145 and 4-2 cell lines, the 4-2 tumor showed a lower expression of the oncogene MUC1 than those found in the DU145 tumor (Figure 5, D and E). Finally, the staining for the CDK4 protein, a cell cycle activator, was drastically lower in the 4-2 tumor when compared with the DU145 tumor (Figure 5, F and G), which is in agreement with the lower proliferation rates of the 4-2 cell line described earlier (Figure 4).

Figure 5.

In vivo tumorigenicity assay and immunohistochemistry. (A) 3.0 x 10⁶ DU145 and 4-2 cells were subcutaneously injected in 4-week-old nu/nu female mice (two injections/mice, five mice/group). The length and the width of the tumors were measured three times per week for 31 days. Values shown are mean ± SEM of tumors volumes for all tumors measured per mice group. Immunohistochemical staining of tumors derived fromDU145 and 4-2 cell lines for p53 (B, C), MUC1 (D, E), CDK4 (F, G), and PSA (H, I) proteins. All images were acquired at the same magnification (100x).

Lower PACE4 Expression Induces Apoptosis in the 4-2 Cell Line

Because PACE4 expression correlated with the reduced cell proliferation of DU145 cells, we decided to analyze the cell cycle of DU145, 4-2, and 4-2 + PACE4 cells by flow cytometric analysis using the PI staining. As shown in Figure 6A, the proportion of hypodiploid cells (characteristic of apoptotic cells) is higher in the 4-2 cell line (12.3%) than those found either in DU145 or in 4-2 + PACE4 cells (1.74% and 3.30%, respectively). Moreover, 78.8% of the nonapoptotic 4-2 cells were blocked into the G₀/G₁ phase of the cell cycle, and only 3.84% were into the G₂-M phase. This is in contrast with the DU145 cells, where only 67.3% of nonapoptotic cells were in the G₀-G₁ phase, and 14.5% entered into the G₂-M phase. However, 14.5% of the latest cells are in the G₂-M phase, whereas only 3.84% of 4-2 cells are distributed in this phase. Thus, the lower cell proliferation observed in the 4-2 cell line can be attributed to a higher apoptosis rate and a blockade into the G₀-G₁ phase of the cell cycle. Interestingly, the study of the partial rescue cell line 4-2 + PACE4 showed that a lower population of these cells were in apoptosis when compared with the 4-2 cells (3.30% and 12.3%, respectively), confirming an important role of PACE4. However, these cells with a partially restored PACE4 expression still seemed to be blocked into the G₀/G₁ phase of cell cycle as for the 4-2 cells (80.6% and 78.8%, respectively) and like 4-2 cells, the 4-2 + PACE4 cells had a low percentage of cells into the G₂-M phase (3.61%). Thus, the partial reestablishment of PACE4 expression levels leads to a lower apoptotic percentage of DU145 cells but still results in a partial blockade of the cell cycle into the G₀/G₁ phase.

Figure 6.

Cell cycle analysis and identification of an apoptotic biomarker by MALDI-TOF/TOF. (A) DNA content of DU145, 4-2 and 4-2 + PACE4 cells was determined by flow cytometric analysis using PI staining without cell fixation. The 4-2 cells showed a higher level of fragmented DNA than DU145 and 4-2 + PACE4 cells (12.30%, 1.74%, and 3.30%, respectively), indicating a higher population of hypodiploid cells for the 4-2 cell line, a characteristic of apoptotic cells. Inserts show the distribution of nonapoptotic cells into the phases of cell cycle. (B and C) Detection and identification of the apoptotic biomarker TRPS1 were done by MALDI-TOF/TOF in positive reflectron mode from different secretomes coming from triplicates of DU145 and 4-2 conditioned culture media after trypsin digestions. B represents the mass spectra comparison between conditioned media from both cell lines and the box shows a biomarker with a m/z of 1228 in 4-2 CM. C shows the identification by fragmentation of this peak as the zinc finger transcription factor TRPS1 by MALDI-TOF/TOF.

Because important differences were observed between DU145 and 4-2 cell lines in regard to apoptosis, we sought potential effects on proapoptotic factors (Figure 6B). On the type of analysis, we used DU145 and 4-2 cell line CM digested with trypsin to perform a MALDI-TOF/TOF analysis. One marker of interest was identified at amass of 1227.8Da only in the 4-2 CM. The protein identification was then performed under the MASCOT sequence query search program (Figure 6B). This biomarker corresponded to a fragment of the TRPS1 transcription factor, a protein issued from the apoptosis-associated gene GC79/TRPS1 that encodes an atypical multitype zinc finger GATA-type transcription factor associated to prostate cancer that potently and specifically represses transcriptional activation mediated by other GATA factors [56,57]. Although there is no evidence for the secreted release of TRPS1 or of a possible isoform, we might conclude that its detection in CM is the result of increased apoptosis in the 4-2 cell line.

Discussion

The relationship between PCs and cancer has become stronger within the last few years. However, because cancer cell lines generally coexpress multiple PCs, it still remains unclear if only one or multiple PCs are more important in cancer progression. In this study, we chose to focus on PACE4 because of our finding that this PC is specifically overexpressed in different clinical stages of prostate cancer tissues, arguing for a particular role of PACE4 in the establishment or the maintenance of prostate cancer (Figure 1). To reinforce our findings in human prostate cancer tissues, we also evaluated six independent studies that also report the selective PACE4 overexpression in prostate cancer (Table 1), confirming the potential importance of this enzyme to sustain the disease.

There are various ways in which we could choose to study the role of PACE4 in prostate cancer. Molecular and pharmacological inhibitions have particular advantages, as long as they can be shown to be specific. At the present time, molecular inhibition (gene knockout or knockdown) is clearly more specific because pharmacological inhibitors of PCs still lack PC specificity (i.e., inhibit two or more PCs with equal potency). In regard to prostate cancer, various accepted human model cell lines have been previously well characterized. Therefore, for this study, we chose to establish the molecular inhibition of PACE4, using SOFA ribozyme technology, an exquisitely specific method of gene silencing, in the androgen-independent prostate cancer cell line DU145. After PACE4-SOFA-HDVRz transfection, a lower-than-expected number of stable cells were obtained, and transfected DU145 cells grew very slowly. In retrospect, based on all the evidence presented herein, this was most likely a direct consequence of lowered PACE4 levels, arguing a role of PACE4 in DU145 cell proliferation.

Considering the high selectivity of PACE4-SOFA-HDVRz, we chose the cell line with the lowest expression level of PACE4 mRNA for further studies (Figure 2C). Northern blots were performed for the other endogenously expressed PCs to illustrate that this effect was specific to PACE4 (Figure 2D). We also attempted to transfect the cell line with the lowest PACE4 levels with an expression vector containing the PACE4 cDNA. We were surprised to observe that PACE4 RNA levels were reestablished because this cell line still expressed the tRNA^Val-PACE4-SOFA-HDVRz (Figure 2C). This suggests that within a cellular context, the turnover rate of the ribozyme was not as high enough to affect a much larger pool of mRNA issued from expression vectors. Nonetheless, this partially reestablished cell line served our purpose as a control. We believed that this was the best control to use because it is directly derived from cells used for all experiments, eliminating all possible cell line-to-cell line variations. However, further improvements of the SOFA-HDVRz will eventually be required to reach higher turnover rates without significantly affecting its specificity.

To confirm that the established 4-2 cell line with lower PACE4 expression level was a suitable model of prostate cancer, we undertook a comparative gene expression profiling study (Figure 3). Among the downregulated genes in the 4-2 cell line, MUC1 is the one whose expression was the most affected by the PACE4 knockdown (Table W1). Mucin 1 is an oncoprotein expressed at high levels in diverse human carcinoma that is implicated in cellular transformation. Indeed, by its interactions with diverse effectors including EGFR, β-catenin, p53, IκN kinase, and NFκB p65, MUC1 blocks the induction of apoptosis and necrosis in human carcinoma [52,53]. A tissue microarray study reported the overexpressed of MUC1 in 58% of human primary prostate cancer tissues [58], whereas a correlation of its expression level in human prostate carcinoma tissues with the most advance stages of the disease was also reported [59]. The androgen-dependent prostate cancer cell line LNCaP does not express MUC1, the androgen-independent PC-3 cells expresses low constitutive levels, whereas DU145 cells highly and constitutively express MUC1 [52,60]. Our study now strongly correlates MUC1 expression with PACE4 expression level, as confirmed by quantitative PCR on the 4-2 cell line (Figure 3B) and immunohistochemical staining of DU145 and 4-2 tumor xenografts (Figure 5, D and E). Interestingly, the inhibition of MUC1 in DU145 cells with a peptide inhibitor interfering with its oligomerization led to similar effects than those reported in the present study for the 4-2 cell line; that is, inhibition of cell proliferation, induction of cell death pathways, and tumor regression of DU145 xenografts growing in nude mice [52].

Some of the effects reported in the present study could be directly linked to the reduced MUC1 expression caused by PACE4 inhibition, arguing for a central role of this overexpressed PCs in prostate carcinoma. Although this hypothesis needs further investigation, the study of MUC1 transcription regulators could lead to a better understanding of PACE4 contribution to its expression control, such as with known transcriptional regulators of MUC1 include TNF-α, IL-6, IFN-γ, TGF-β1, and HIF1-α [53]. In addition to TGF-β1, a known PC substrate [26], PACE4 could also regulate MUC1 expression indirectly by activating metalloproteinases such as ADAM17/TACE [61], which is directly involved into the regulation of TNF-β and IL-6 receptor [3]. Thus, considering the broad spectrum of potential endogenous substrates of PACE4, further investigations of its direct or indirect actions on the transcriptional control of these genes will be needed to better understand the gene expression profile of 4-2 cells.

The consequences of lowered levels of PACE4 are well illustrated by the reduced cell proliferation rate and the incapacity of these cells to form subcutaneous tumors in nude mice (Figures 4 and 5, respectively). The partial restoration of PACE4 expression levels in this cell line allowed a partial recovery of the in vitro proliferation rate, suggesting that PACE4 is a key player for tumoral growth and its expression levels have to be high to achieve this function (Figure 4). These results are in great agreement with the reduced levels of the CDK4 protein in the tumor derived from the 4-2 cell line compared with those found in the tumor derived from the DU145 cell line (Figure 5, F and G) because this protein is known to control the cell cycle. Moreover, analysis of the cell cycle in the DU145, 4-2 and 4-2 + PACE4 cell lines revealed the importance of PACE4 in regulating apoptosis and the cell cycle (Figure 6A). These results also indicated that partial recovery of PACE4 expression blocked DU145 apoptosis but could not reverse the blockade of cells into the G₀-G₁ phase, suggesting that full PACE4 activity is necessary to sustain the high proliferative state of DU145 cells.

In an attempt to identify other biomarkers affected by PACE4 in the proliferation of DU145 cells, we undertook proteomic analyses of DU145 and 4-2 cells derived conditioned media (Figure 6, B and C). In the samples issued from 4-2 cells, a fragment derived from the tryptic digestion of TRPS1 (a multitype zinc finger GATA-type transcription factor) was consistently detected [56]. This transcriptional repressor, whose expression is regulated by androgens through the androgen receptor, is normally expressed in the androgen-dependent LNCaP prostate cancer cells [62,63]. Its lack of expression has been previously reported for the androgen-independent DU145 prostate cancer cells [62], but the stable expression of recombinant TRPS1 in those cells allowed identification of both downregulated antiapoptotic protein POR2 and upregulated proapoptotic protein KAD2 [64]. Our data demonstrate that PACE4 can regulate the expression of the proapoptotic factor TRPS1. Further studies will be required to better characterize the involvement of PACE4 into the transcriptional control of the TRPS1 transcriptional repressor. As mentioned earlier, TRPS1 transcription is known to be regulated by androgens in cells LNCaP prostate cancer cells but not in the DU145 cell line, whose growth is independent of androgens and does not express the androgen receptor [62]. Recently, the bone morphogenic protein 7 (BMP-7), a growth factor belonging to the family of TGF-β that requires a PC cleavage for its propeptide removal [65], has been identified as an inducer of TRPS1 gene expression in mouse kidneys [66]. Thus, further characterization of the TRPS1 genetic regulation by BMP-7 in DU145, or other members of the TGF-β growth factor family that also need furin-like cleavages for processing [3], would be indicated to better understand the cellular functions of the overexpressed PACE4 in human prostate cancer cells.

The genes known to be transcriptionally regulated by TRPS1 include PSA [63], STAT3 [67], parathyroid hormone-related protein (PTHrP) [68], and osteocalcin [69]. Interestingly, the cleavage of human PTHrP and osteocalcin propeptides, essential for the generation of their mature forms, occurs at typical PCs processing sites (₃₁SRRLKR↓AV₃₈ and ₄₆VKRPRR↓YL₅₁, respectively). Thus, the implication of PACE4 catalytic activity to the transcriptional control of TRPS1 and DU145 apoptosis triggering could be through these potential substrates. This hypothesis is supported by our genomic analysis, where the gene PTHLH (coding for PTHrP) was upregulated in 4-2 cells (Table W2).

Regarding the potential therapeutic avenues targeting PACE4, these include various approaches such as small molecule inhibitors [28], monoclonal antibodies [70], or even ribozymes or siRNA complexed with liposomes [71]. Small molecule inhibitors may be problematic because none of these have been shown to be specific to a particular PC. This is due to the high similarity of the catalytic regions of each PC. Whereas monoclonal antibodies could provide better specificity, it is not clear whether PACE4 localized at the cell surface and/or intracellular PACE4 will require targeting [72]. Finally, ribozyme or siRNA complexed with liposomes would certainly be highly specific but require additional research to make these reagents druggable. One potential avenue of research that offers both specificity and cell compatible characteristics (i.e., little toxicity and cell permeability) is the development of small peptide inhibitors (∼1000 Da) targeting PACE4 [9]. Combined with peptidomimetics, such studies could lead to the development of novel drugs for prostate cancer.

In conclusion, this study reports the specific overexpression of PACE4 in prostate cancer tissues. Our results showed that PACE4 plays a major role in the progression of prostate tumors to a higher aggressive status, and its inhibition leads to a significant loss of tumorigenicity of the prostate tumor cell line DU145. A better understanding of PACE4 direct and indirect cellular targets contributing to these phenomena remains a great challenge and implicates multiple known oncogenic factors. Nonetheless, the data strongly validate PACE4 as an important therapeutic target in human prostate cancer.