Prostate cancer cell lines LNCap, 22RV1, PC-3 and DU145 were obtained from the ATCC (American Type Culture Collection, Manassas, VA, USA) and cultured in RPMI-1640 medium (Gibco-BRL, Invitrogen Corp., Grand Island, NY, USA) with 1 mM sodium pyruvate, 100 U/ml penicillin G, 100 μg/ml streptomycin (Sigma Corp. of America, Ronkonkoma, NY, USA), 10% fetal bovine serum (FBS; Tianjin Hao Yang Biological Manufacture Co., Ltd., Beijing, China). Lipofectamine™ 2000 transfection reagent, TRIzol and the One-Step Fluorescence assay kit were products of Invitrogen (Carlsbad, CA, USA).

Prostate tissue samples including normal prostate (n=12), primary prostate cancer (n=47), and benign prostatic hypertrophy (BPH) (n=26), were obtained at the Department of Urology, the First Affiliated Hospital of Sun Yat-sen University, Guangzhou, China. All tissue samples were stored at −80°C until being further processed. The Gleason score was used to evaluate tumor malignancy, and the prostate cancer samples were then divided into two categories according to their Gleason score (Table I). Informed content for the scientific use of biologic materials was obtained from all patients in accordance with the requirements of the Medical Ethics Review Board of Sun Yat-sen University.

One step TaqMan real-time RT-PCR was used to evaluate the PSMA silencing effect. Specific probes and primers of PSMA were designed by Beacon Designer 7.01 Demo, and the β-actin gene served as the internal control. Primer and probe sequences are shown in Table II. One step real-time PCR amplification reaction of PSMA contained 1 μl SuperScript III RT/Platinum Taq Mix, 1 μl RNaseOut, 25 μl 2X reaction mixture (Invitrogen), 0.5 μM of the forward and reverse primers and 0.5 μM of the probe, and 3 μl of RNA and nuclease-free water to a final volume of 50 μl. Real-time RT-PCR was performed on an ABI 7500 real-time PCR system (Applied Biosystems, Foster City, CA, USA) using thermal cycling conditions as follows: 50°C for 15 min, 95°C for 2 min; 95°C for 15 sec, 60°C for 30 sec, 40 cycles. All PCR reactions were performed in duplicate.

Reverse transcription (RT) and real-time PCR based on SYBR-Green detection was used to identify candidate PSMA-related genes in PCa cell lines and tissues. Total RNA was extracted by TRIzol reagent following the manufacturer’s instructions. Briefly, PCa cell total RNA was extracted directly in a 3.5 cm diameter culture dish by adding 1 ml TRIzol reagent, and the total RNA of the PCa clinical samples that were cut into ~0.5 cm pieces, adding the appropriate volume of pre-cooling TRIzol regeant at v/v 10:1. RT reaction was carried out in a final volume of 20 μl using the Invitrogen reverse transcription kit (Invitrogen) according to the manufacturer’s instructions. First, the reaction mixture containing 500 ng/μl Oligo (dT)₁₈ (1 μl), total RNA (1.5 μg), 10 mM each deoxynucleoside triphosphate (dNTP) (1 μl) and nuclease-free water up to a volume of 13 μl was incubated for 5 min at 65°C, placed on ice for more than 1 min, then 5X first strand buffer (4 μl), 0.1 M DTT (1 μl), RNase inhibitor (1 μl), 200 U/μl SuperScript III reverse transcriptase (1 μl) were added to a final reaction volume of 20 μl. The thermal cycling parameters for the RT reaction were as follows: 60 min at 50°C and 15 min at 70°C. Products of RT reactions were stored at −20°C until further analysis. The mRNA expression of candidate genes was determined using the ABI PRISM 7500 detection system (Applied Biosystems) and SYBR Premix Taq™ (Invitrogen). The primer sequences are shown in Table II. The reaction system contained 2X SYBR Premix Ex Taq™ II (10 μl), 0.4 μM of each forward and reverse primers, 50X ROX reference dye 0.4 μl and 2 μl of the synthesized cDNA in a 20 μl reaction volume. Thermal cycling conditions were as follows: 95°C for 30 sec, followed by 40 cycles of 95°C for 5 sec, 60°C for 30 sec. All PCR reactions were performed in duplicate and repeated three times. Melting curve analysis was used to control the amplification specificity. Target gene expression is shown as the fold increase or decrease relative to the expression of β-actin. Fold-changes in target gene mRNA expression were determined as 2^−ΔΔCt using the same calculation formula as determined in the microarray analysis.

Three pairs of siRNAs for PSMA (GenBank ID: M99487) were designed by online software (http://jura.wi.mit.edu/bioc/siRNAext/siRNA_search.cgi?tasto=10925811) (Table III). Designed siRNAs and the positive and negative control siRNAs were synthesized and supplied by Guangzhou RiboBio Co., Ltd. (Guangzhou, China). The day before transfection, LNCap cells were collected, digested with 0.25% trypsin and counted, and then seeded onto 6-well plates (Corning Incorporated Costar, Corning, NY, USA) at the initial density of 2×10⁵ cells/well and cultured with antibiotic-free medium. The transfection was performed by Lipofectamine™ 2000 with the concentration of siRNA at 50 nM, and LNCap cells transfected with the negative control siRNA (supplemented by RiboBio) served as the control. Total protein and total RNA were extracted 72 h after transfection. One step real-time RT-PCR was used to evaluate the PSMA silencing effect at the mRNA level. Silencing effect of PSMA at the protein level was identified by western blotting. The monoclonal antibody (YPSMA, ab19071; Abcam Inc., Cambridge, MA, USA) used in the western blotting recognized 716–723 amino acid residues of the C terminus of PSMA. The antibody dilution ratio used in the western blotting was 1:1,000.

Human tumor metastasis PCR array (SABiosciences, Qiagen, Gaithersburg, MD, USA) was designed to analyze 96 genes by real-time PCR, including 84 genes known to be involved in tumor metastasis, 4 housekeeping genes (HK genes) and 8 quality control genes. Tumor metastasis-related genes selected for this array encode several classes of protein factors including cell adhesion, ECM components, cell cycle, cell growth and proliferation, apoptosis, transcription factors and regulators and other genes related to tumor metastasis.

Two groups were designed for the tumor metastasis PCR array analysis: the PSMA(−) group was used to analyze the differentially expressed genes between the LNCap and PC-3 cells, and the si-PSMA group was used to analyze the differentially expressed genes between the si-PSMA LNCap and LNCap cells. The mean value of the replicates for each group was calculated and expressed as the cycle threshold (Ct). The amount of gene expression was then calculated as the difference (ΔCt) between the Ct value obtained for the target gene and the Ct value obtained for the HK genes. The value of ΔCt was calculated for each pathway-focused gene in each group using the following calculation formula: ΔCt (control group) = average Ct of the target gene − average Ct of the HK genes for the control group array; ΔCt (experimental group) = average Ct of the target gene − average Ct of the HK genes for the experimental group array. The ΔΔCt was calculated for each gene across two PCR arrays (or groups), ΔΔCt = ΔCt (experimental group) − ΔCt (control group). The fold-change for each gene from the control group to the experimental group was calculated as 2^−ΔΔCt.

Based on the comparison of the common genes between the PSMA(−) group and the si-PSMA group, and the previous studies and reports, the MMP3, MTSS1 and CDH6 genes were chosen and further analyzed by reverse transcription (RT) and SYBR-Green-based real-time PCR in PCa cell lines including the PSMA-positive cell lines (LNCap and 22RV1) and the PSMA-negative cell lines (PC-3 and DU145), and also 85 clinical samples of prostate tissues, including 12 normal prostate, 26 BPH and 47 prostate cancer samples. The pathological information regarding the prostate tissues is shown in Table I.

The PSMA mRNA expression levels were not distributed normally, therefore the log-transformed relative expression levels were used to obtain sufficiently normally distributed values. The differences in expression of MMP3, MTSS1 and CDH6 genes among normal prostate, BPH and PCa were evaluated using one-way ANOVA. Spearman correlation coefficient was used to evaluate the correlation between Gleason score and the levels of MMP3, MTSS1, CDH6 and PSMA mRNA expression, and also between PSMA and MMP3, MTSS1 and CDH6. P-values <0.05 were considered to indicate a statistically significant result. All statistical analyses were performed using SPSS11.5 software (SPSS Inc., Chicago, IL, USA).

Stable PSMA silenced LNCap cell lines (si-PSMA LNCap cells) were successfully obtained by transfection with the siRNA sequences. Seventy-two hours after transfection with the siRNAs, the PSMA gene expression level was detected by western blotting and real-time PCR. Real-time PCR showed that the PSMA mRNA expression level was downregulated >4-fold by siRNA-3, and the western blot analysis showed that the PSMA protein expression level was reduced 68% by siRNA-3 (Chi-square test, P<0.01 compared to the negative control; Fig. 1). siRNA-1 and siRNA-2 exhibited a weaker silencing effect than siRNA-3, therefore, siRNA-3 was chosen for further study.

In order to gain better understanding of the molecular mechanism involved in the metastatic progression of androgen-independent prostate cancer cells, we performed a comprehensive metastasis-related gene expression profile analysis of PSMA-positive LNCap and PSMA-negative PC-3 cells, namely the PSMA(−) group, using a tumor metastasis PCR array. Genes whose expression was upregulated or downregulated >2-fold, were considered significantly differentially expressed and were selected for further analysis. Forty-one genes had significantly higher expression in the PC-3 cells (P<0.05; Fig. 2), among which 23 genes were expressed >10-fold (Table IV). The genes with the highest upregulated expression in the PC-3 cells were CDH11 (87,443.2-fold). There were 15 genes with downregulated expression in the PC-3 cells when compared to the LNCap cells; among which the expression levels of 6 genes (TP53, PTEN, CTNNA1, CDH1, MDM2 and IGF1) were >10-fold (Table IV). Most of the significant genes identified in the present study have not been previously reported in prostate cancer.

We further analyzed the differential expression of tumor metastasis-related genes in the si-PSMA LNCap cells in which PSMA expression was blocked. Tumor metastasis PCR array analysis revealed that 10 genes were upregulated (namely CDH6, CXCL12, IL18, IL8RB, IGF1, MMP13, MMP3, MTSS1, TSHR and ITGA7) and 4 genes were downregulated (CCL7, ITGB3, MDM2 and MMP2) significantly in the si-PSMA LNCap cells when compared to the PSMA non-silenced LNCap cells. The differentially expressed genes and their description are documented in Table V. Consistent with other reports, the IGF1 gene has been previously reported to be upregulated in C4-2 cells when compared with LNCap cells (18). Among the total 14 genes that were significantly altered in the si-PSMA LNCap cells, the MDM2 gene showed the maximum downregulated change when PSMA expression was suppressed (Table V), consistent with a previous report (19).

By comparison of the differentially expressed genes between the PSMA(−) group and the si-PSMA group, we aimed to identify those common genes that may be regulated by or related to PSMA. From the tumor metastasis PCR array results, we found that expression of 6 genes (namely IL18, MTSS1, MMP13, ITGA7, MMP3 and CDH6) was significantly upregulated in both the si-PSMA LNCap and the PC-3 cells, although their expression levels were different. Expression of 2 genes (MDM2 and MMP2) was downregulated to a similar extent in both the PC-3 and the si-PSMA LNCap cells. The results indicate that these 8 genes may be related to the role of PSMA in tumor metastasis.

The genes differently expressed between the PSMA(−) and si-PSMA groups included 2 genes (CCL7 and ITGB3) that were upregulated significantly in the si-PSMA LNCap cells but not in the PC-3 cells. Expression of 13 genes was significantly lower in the PC-3 cells but not in the si-PSMA LNCap cells. Moreover, expression of IGF1 and CXCL12 was significantly upregulated in si-PSMA LNCap cells, but significantly downregulated in the PC-3 cells. The ΔCt changes in 28 genes were <2 and showed no significant change in both the PC-3 cells and the si-PSMA LNCap cells.

Three candidate genes, namely MMP3, CDH6 and MTSS1, were chosen for further analysis and identification in the PCa cell lines and prostate tissues by real-time RT-PCR. PCa cell lines used for verification included the PSMA-positive LNCap and 22RV1 cells, and the PSMA-negative PC-3 and DU145 cells. Consistent with the results of the tumor metastasis PCR array analysis based on LNCap and PC-3 cells, the MMP3 mRNA expression level was downregulated in the PSMA-positive cells including LNCap and 22RV1, and upregulated in the PSMA-negative cells including PC-3 and DU145 (P<0.05; Fig. 3A). CDH6 and MTSS1 levels were significantly higher in the PC-3 cells when compared with the other 3 cell lines (P<0.05; Fig. 3A). Real-time PCR confirmed the microarray results for MMP3, CDH6 and MTSS1 in the LNCap and PC-3 cells, and the PSMA-positive 22RV1 cells showed the same differential expression trend as that in LNCap cells. In contrast, the expression trend of CDH6 and MTSS1 in the PSMA-negative DU145 cells was different from that in the PC-3 cells.

The results of real-time RT-PCR identification in clinical samples showed that the relative MMP3, CDH6 and MTSS1 expression level in normal prostate, BPH and PCa tissues decreased with the development of prostate cancer, and showed a contrary trend with the expression level of PSMA. Expression levels of MMP3, MTSS1 and CDH6 were significantly lower in PCa compared with BPH and normal prostate, respectively (P<0.05). Correlation analysis demonstrated a negative correlation between the MMP3, MTSS1 and CDH6 mRNA expression and Gleason score (for MMP3, r=−0.5826, P<0.0001; for MTSS1, r=−0.4727, P<0.0001; for CDH6, r=−0.4774, P<0.0001; Fig. 3B). Negative correlation was also observed between the levels of PSMA and MMP3, MTSS1 and CDH6, indicating that these genes may be involved in the regulation of prostate cancer metastasis mediated by PSMA (for MMP3, r=−0.6811, P<0.05; for CDH6, r=−0.5430, P<0.05; for MTSS1, r=−0.5540, P<0.05) (Fig. 3B).