Genetic and cellular studies highlight that A Disintegrin and Metalloproteinase 19 is a protective biomarker in human prostate cancer

ADAM19 immunohistochemistry was conducted on human prostate cancer samples contained on the Prostate Cancer Tissue Array (Abcam, #ab178263). We personally did not have to gain ethics approval as samples were part of a commercially available tissue array. All tissue was examined/diagnosed by a licensed pathologist and was ethically obtained. Immunohistochemistry was conducted using standard procedures with primary antibody (rabbit anti-hADAM19 disintegrin domain IgG (pAb362)) at a 1:200 dilution [22, 23].

A human prostate cancer microarray of 71 patients (GEO accession number: GSE40272) contained information on ADAM19 gene expression in human prostate tumours, and was processed using the R ‘affy’ and ‘limma’ packages. In addition, we investigated the clinical significance of human ADAM19 expression in human prostate cancer tumour tissue in this cohort of patients, as follow up clinical data was available.

We also analysed intratumoural RNA-seq expression data from a cohort of 156 patients with prostate cancer available at The Cancer Genome Atlas (TCGA) (http://​tcga-data.​nci.​nih.​gov/​tcga/​tcgaDownload.​jsp); accessed June 2013). This cohort consisted of 65 patients with pathologically determined stage II prostate cancer, 85 patients with stage III, 5 patients with stage IV prostate cancer and one patient of unknown staging. The mean age of patients in this cohort was 60.3 years.

We personally did not have to gain ethics approval as analysis was performed on publicly available microarray data. The Cancer Genome Atlas (TCGA) is advised by an External Scientific Committee whose membership includes patient advocates, senior scientists and clinicians with relevant expertise in ethics. All prostate samples used in the GSE40272 related study were collected with patient’s informed consent under an Institutional Review Board approved protocol.

Normal human epithelial prostate cells (RWPE-1), which express the androgen receptor, were compared with androgen sensitive, human prostate cancer cells (LNCaP). Androgen independent human prostate cancer cells (PC3) were used for in vitro scratch assays due to their ability to produce a monolayer in culture. Human embryonic kidney cells (HEK293) were used for tumor necrosis factor-α (TNF-α) shedding experiments.

All cells were purchased from the American Type Culture Collection (Manassas, VA, USA). HEK293 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) [low glucose; Gibco] containing 10 % fetal calf serum (FCS) and 1 % penicillin/streptomycin (Invitrogen, USA). RWPE-1 cells were cultured in Keratinocyte Serum Free Medium (K-SFM; GIBCO) containing 0.05 mg/ml bovine pituitary extract (BPE) and 5 ng/ml human recombinant epidermal growth factor (EGF) provided with the K-SFM kit. LNCaP and PC3 cells were cultured in Roswell Park Memorial Institute-1640 media (RPMI) (Sigma-Aldrich, Germany) with 10 % FCS and 1 % Penicillin/Streptomycin. To maintain viable healthy and undifferentiated cells, RWPE-1, LNCaP and PC3 cells were maintained until they reached 70 % confluency and were then transferred into a 75 cm² flask. Cells were split into 6 well Cell Bind (Costar), 12 well Cell Bind (Costar) or 96 well cell culture plates for further studies.

LNCaP and RWPE-1 cells were harvested and washed with cold 1X PBS. Cells were lysed using cytosolic extraction buffer (10 mM hydroxyethyl piperazineethanesulfonic acid; 3 mM MgCl₂; 14 mM KCl; 5 % glycerol; 0.2 % IGEPAL) containing phosphatase and protease inhibitors (Roche). Cells were then scraped and lysates were transferred to a 1.5 mL eppendorf tube and stored at -80 °C. After 24 h, lysates were centrifuged at 13 000 rpm at 4 °C for 10 min. Bradford assay (Bio-Rad, Hercules, CA, USA) was used to determine protein concentrations. Protein lysates (40 μg) were solubilized in Laemmeli sample buffer and boiled for 10 min, resolved by sodium dodecyl sulfate (SDS)–polyacrylamide gel electrophoresis on 10 % polyacrylamide gels, transferred by semi-dry transfer to polyvinylidene difluoride membrane and blocked with 5 % milk powder. Membranes were then incubated overnight at 4 °C in primary antibodies [rabbit anti-hADAM19 metalloproteinase domain IgG (pAb361) [22] or mouse anti-β-actin (Abcam, Cambridge, UK; ab6276)] using recommended dilutions. Membranes were washed three times in washing buffer and incubated for 60 min at room temperature with either anti-rabbit or anti-mouse horse-radish peroxidase (HRP; Sigma, USA) respectively. Membranes were then washed and briefly incubated in Amersham ECL Prime Western Blotting Detection Reagent (GE). The protein bands were detected using the Alpha Innotech MultiImage II Fluor Chem FC2.

Transfections were conducted in 6 or 12 well Cell Bind (Costar) culture plates. Transfection was carried out once adherent HEK293 or PC-3 cells reached approximately 70 % confluency using Lipofectamine™ 2000 (Invitrogen, Calsbad, California, USA). Alternatively, LNCaP cells were transfected in suspension. Cells were transfected with either pcDNA3.1 GFP vector (Invitrogen), empty pCR3.1 vector [23] or vectors containing the cDNA for human ADAM19 (pCR3.1 hADAM19) [23] or human TNF-α (pcDNA3.1 (-) pro-TNF-α) [24]. Cells were incubated at 37 °C, in 95 % O₂/5 % CO₂. Cells were visualised for GFP using the Nikon Eclipse Ti microscope to evaluate transfection efficiency after 24 and 48 h. Cell-free culture supernatants were collected after 48 h. Transfected cells were then used for immunocytochemistry to evaluate ADAM19 expression. In addition, ADAM19 transfected cells were used in MTS proliferation assays or migration studies. Empty vector-transfected cells were used as a comparative control.

Immunocytochemistry was used to confirm basal level and over-expression of human ADAM19 in LNCaP, RWPE-1, PC-3 and HEK293 cell lines. Cells were fixed in methanol/acetone (1:1) and endogenous peroxidases blocked using 0.3 % hydrogen peroxide in Triton X/PBS (Tx/PBS) for 5 min. Cells were blocked for 1 h in 10 % FCS/Tx/PBS, incubated with primary antibody (rabbit anti-hADAM19 disintegrin domain IgG (pAb362) [23]) at 4 °C overnight, washed 3X in Tx/PBS for 5 min before a secondary antibody [anti-rabbit horse-radish peroxidase (HRP) (Sigma, USA) diluted 1:100 in blocking buffer (10 % FCS/Tx/PBS)] was added for 45 min. Cells were washed 2X in Tx/PBS for 5 min before diaminobenzidine (DAB, DAKO) was added for approximately 10 min and cells were then visualised. Negative controls had the primary antibody omitted which resulted in no staining. Cells were visualised using the Nikon Eclipse Ti microscope.

Transfected and untransfected LNCaP cells were resuspended at 0.25x10⁵ cells/mL in RPMI-1640 medium containing 10 % FCS and 1 % streptomycin/ penicillin and added in 100 μl volumes to the centre of the wells of a 96-well culture plate. This technique allowed even dispersion of cells in the well. We plated 12 samples per cell type per treatment per time point. After 1, 3, 5 and 7 days, the medium was carefully aspirated and 100 μl RPMI-1640 medium containing 10 % FCS and 1 % streptomycin penicillin containing 20 μl of MTS assay reagent was added to each well for 3 h. After incubation at 37 °C, in 95 % O₂/5 % CO₂, proliferation was determined by MTS assay. Plates were read at 490 nm (0.1 s per well) on a plate reader. The Nikon Eclipse Ti microscope was used at each required time point to image cells.

Human TNF-α in the cell-free culture supernatant collected from transfected HEK293 or PC3 cells was measured using a commercially available enzyme-linked immunosorbent assay kit (TNF-α; R&D Systems, DY210).

Migration of transfected PC3 cells was assessed using an in vitro scratch assay [25]. Cell death was determined with trypan blue cell counting.

Statistical analysis of microarrays was performed using the R programming environment. The Kaplan-Meier survival curve was based on unadjusted Cox regression of GSE40272 data using the R “survival” package. The median was used to divide tumours into high or low intratumoral ADAM19 expressing groups for comparison with longitudinal survival.

The TCGA boxplot was produced using the R “graphics” package and showed the correlation between tumour stage and human ADAM19 expression. This data was additionally assessed using Pearson’s product moment correlation. Statistician Dr. Patrick Candy performed the microarray statistical analysis (Harry Perkins Institute of Medical Research, University of Western Australia, Australia).

Human prostate tumour biopsies and normal prostate tissue samples on a Prostate Cancer Tissue Array were immunostained for ADAM19. In normal human prostate tissue, ADAM19 is highly expressed on the luminal surface of glandular epithelial cells as indicated by brown diaminobenzidene staining (Fig. 1a). Excitingly, we report for the first time that human prostate carcinoma samples have low ADAM19 expression (Fig. 1c and d) when compared with benign prostate hyperplasia samples (BPH; Fig. 1a and b). Intriguingly, human ADAM19 expression is reduced as the severity of prostate cancer rises (Fig. 1c and d) which is a novel finding.

In order to evaluate the relationship between ADAM19 levels and prostate cancer, we studied ADAM19 expression in publicly available microarray data from two distinct cohorts of prostate cancer patients. In the GSE40272 cohort (Fig. 2a), there was a significant association between high median ADAM19 expression levels and reduced cancer relapse (Hazard Ratio 0.1749, p < 0.003). The clinicopathological characteristics of the GSE40272 cohort is presented in Table 1.

In the TCGA cohort (Fig. 2b) we found that high ADAM19 expression in prostate cancer tissue was significantly negatively associated with tumour stage (cor = -0.18, p < 0.026). There were few deaths in the TCGA cohort, preventing any meaningful association of ADAM19 expression to overall survival. However, the TCGA cohort showed that high ADAM19 expression was highly associated with lower tumour stage, which taken together with the strong association of high ADAM19 expression with higher disease free survival, provides substantial evidence that ADAM19 is a marker of improved prognosis in prostate cancer.

Having demonstrated that high levels of ADAM19 mRNA expression correlate with increased disease free survival and lower tumour stage in publicly available prostate cancer microarray databases, we then sought to determine the ADAM19 expression levels in human tumorigenic LNCaP prostate cancer cells and normal RWPE-1 prostate epithelial cells.

To ensure that our cells were displaying expected proliferative capacity, we conducted proliferation assays with LNCaP and RWPE-1 cells. As expected, we found that LNCaP cells proliferated significantly faster (p < 0.05) than normal RWPE-1 prostate epithelial cells at 1, 3 and 5 days post-seeding (Additional file 1: Figure S1).

We next investigated the level of expression of ADAM19 protein in LNCaP cells and normal RWPE-1 cells. As shown in Fig. 3a, human ADAM19 is endogenously expressed in both LNCaP and RWPE-1 cells, but to different degrees. The 80 and 45 kD bands were observed at considerably higher levels in RWPE-1 cells, consistent with the notion that ADAM19 may act as a tumor suppressor in prostate cancer cells. Furthermore, RWPE-1 cells showed higher expression of endogenous ADAM19 protein in immunocytochemistry experiments (Fig. 3b). In addition, we found the intracellular expression of ADAM19 was cytoplasmic and heterogenous in both LNCaP and RWPE-1 cells. Taken together, these data suggest that normal prostate cells express significantly higher levels of ADAM19 compared to their tumorigenic counterparts.

To determine the bioactivity of ADAM19 in our over-expression system, we conducted transfections in HEK293 cells. We showed that ADAM19 is readily over-expressed in HEK293 cells and that it is predominantly cytoplasmic and heterogeneous in distribution (Additional file 1: Figure S2A). Transfection efficiency, as determined using a GFP vector, was greater than 70 % (Additional file 1: Figure S3). Next we validated that we could use shedding of the pro-inflammatory cytokine TNF-α, a known substrate for ADAM19 [26, 27], as a bioassay for ADAM19 activity. Co-transfection of vectors expressing human TNF-α and human ADAM19 in HEK293 cells resulted in significantly increased TNF-α shedding (17-fold; p < 0.0001) (Additional file 1: Figure S2B). This confirmed in our system that ADAM19 can induce TNF-α cleavage, and generated a simple and reliable tool of its function. As PC3 prostate cancer cells are known to express TNF-α at the mRNA level, we also aimed to assess whether over-expression of ADAM19 in these cells may promote shedding of endogenous TNF-α. Unfortunately, no TNF-α protein was detected in cell-free culture supernatants after ADAM19 over-expression in PC3 cells (data not shown).

We then aimed to ascertain the direct effects of ADAM19 over-expression on human prostate cancer cell proliferation. LNCaP cells were an ideal cell line for these studies, given their lower endogenous levels of ADAM19 (Fig. 3a and b). As shown in Fig. 4a, we were able to effectively overexpress ADAM19 in LNCaP cells, evidenced by strong cytoplasmic diaminobenzidine staining. Transfection efficiency, as assessed with a green fluorescent protein (GFP) expression vector, was more than 50 % (Additional file 1: Figure S4).

The effect of ADAM19 over-expression on LNCaP cell proliferation was determined by MTS assays performed 3, 5 and 7 days after transfection. Figure 4b shows that the proliferation rate of LNCaP cells overexpressing ADAM19 is significantly slower than LNCaP cells expressing empty vector (p < 0.05). Photomicrographs of LNCaP cells taken 7 days post-transfection (Fig. 4c) reinforce the LNCaP MTS proliferation assay results. Taken together, these data suggest that over-expression of human ADAM19 in LNCaP cells reduces proliferation.

Lastly, we examined the impact of ADAM19 over-expression on human prostate cancer cell migration. Androgen-independent PC3 cells were utilised because of their low endogenous expression of ADAM19 (Fig. 5a) and ability to consistently grow in a monolayer. The effect of ADAM19 over-expression on PC3 cell migration was evaluated with an in vitro scratch assay conducted on transfected PC3 cells. We show for the first time that ADAM19 over-expression hinders migration of PC3 cells reproducibly compared with cells transfected with empty vector, 24 h post-transfection (Fig. 5b). This migration pattern was also observed 48 h after the initial transfection (data not shown). In addition, we assessed the cellular viability of PC3 cells 48 h after transfection. We found that ADAM19 transfected PC3 cells experienced statistically significant higher cell death than empty vector transfected cells (Fig. 5c) which is a novel discovery.