Metalloproteome of the Prostate: Carcinoma Cell Line DU-145 in Comparison to Healthy Rat Tissue

Materials and Methods

Animal experiments. Male rats of the Wistar strain Rattus norvegicus (Charles River, Sulzfeld, Germany) were used for these experiments. To attain a selenium-deficient phenotype, one group was fed with a torula yeast-based diet containing 5-10 μg kg^–1 selenium while the control group was fed the same diet but with a sufficient selenium content of about 300 μg kg^–1 added as sodium selenite (MP Biomedicals, Solon, OH, USA). Detailed information about this animal model can be learned from (11). The animals were killed by cardiac puncture under appropriate anaesthesia at the age of 6 months and prostate was obtained from both groups.

Cell culture. The human prostate carcinoma cell line DU-145 (DSMZ, Braunschweig, Germany) was cultivated in RPMI 1640 tissue culture medium supplemented with 10 % fetal calf serum, 100 u ml^–1 penicillin, 100 μg ml^–1 streptomycin and L-glutamine (all Invitrogen, Paisley, GB) at 37°C under 95 % air/5% CO₂ atmosphere.

Sample preparation. Frozen rat prostates were homogenised in a Tris-HCl buffer (25 mmol L^–1, pH 7.4) and harvested DU-145 cell pellet was lysed in a buffer containing 20 mmol l^–1 Tris-HCl; 137 mmol l^–1 NaCl; 10% glycerol (w/v) and 1% nonidet P 40 (w/v) at pH 7.4. To avoid protein degradation all samples were treated at 4°C and EDTA-free protease inhibitor cocktail (Roche, Mannheim, Germany) was added to both buffers.

Homogenates were centrifuged prior to analysis at 186,000 ×g for 1.5 hours and cytotsolic supernatant was used as sample.

Two-dimensional electrophoresis. For two-dimensional electrophoresis (2-DE), proteins were solubilised by adding urea (final concentration: 7.0 mol l^–1) and thiourea (2.0 mol l^–1). Solubilised samples were treated with dithiothreitol (DTT, 70 mmol l^–1) and 4 vinylpyridine (4-VP; 20 mmol l^–1) to reduce and alkylate proteins. Ampholytes (Servalyt, pH=3-10; Serva, Heidelberg, Germany) were added to a final concentration of 2%. Isoelectric focusing on gel pipes as the first (12) and SDS-PAGE (13) as the second dimension were used to separate 80 μg of protein. After separation, proteins were detected by silver staining according to the method of Mortz et al. (14).

Determination of the antioxidative capacity. The antioxidative capacity was determined by a modified ABTS method. In brief, the reagent solution containing the green cationic radical of 2,2’ azino di(3-ethylbenzthiazoline sulphonate) (ABTS^·+) was produced by reaction with ammonium peroxydisulfate (APS) overnight at room temperature. The reagent was then mixed with the samples, degradation of the radical was monitored immediately at 736 nm for 30 minutes in a microplate photometer (Paradigm; Beckman Coulter, Fullerton, CA, USA) and the area under the curve was calculated for each sample. Quantitative evaluation was performed using ascorbic acid standards and results were expressed as mmol ascorbic acid equivalents per g protein.

SEC-ICP/MS. The water-soluble fraction was prepared and the proteins were separated by SEC using an Agilent 1100 HPLC system with a Superdex 75 PC column (3.2 mm × 300 mm; Amersham-Pharmacia, Uppsala, Sweden) and a GFC 3000 guard-column (4 mm × 3 mm; Phenomenex, Torrance, CA, USA). The metal content was determined on-line by ICP-MS (7500c; Agilent Technologies, Santa Clara, CA, USA). An internal standard of 10 μg l^–1 Ir, Rh, Y was added via a T-piece at 50 μl min^–1 between the column and the plasma. Spectra shown here were related to the obtained ¹⁰³Rh signal.

Western-blotting and immunochemistry. Prior immunochemical detection 25 μg protein per sample were separated by SDS-PAGE (13) and transferred onto a nitrocellulose membrane in a wet-blotting system (BioStep, Jahnsdorf, Germany). Blots were blocked in 5% dry milk (m/v) in TBS+T and incubated with primary antibodies for 4 or 7 days at 4°C. Secondary antibodies were labelled with a fluorescent dye and were detected using an infrared imaging system (Odyssey; LiCor, Bad Homburg, Germany). Equal loading of the gels was confirmed by Coomassie-staining of the blots.

Primary antibody directed against thioredoxin reductase was obtained from Abcam (Cambridge, UK) and diluted 1:2000; that against copper and zinc binding SOD was from BioGenes (Berlin, Germany) and diluted 1:1000, and antibody against metallothionein was purchased at Biomol (Hamburg, Germany) and applied at a 1:2000 dilution.

Results

Recent studies demonstrated that many metallo- and metalloid proteins play critical roles in cancer. For instance, a polymorphism (V16A) of the gene coding the manganese-containing form of SOD resulting in higher activity of the enzyme was associated with an increased prostate cancer risk at least in individuals with a low antioxidant status (15, 16). Further examples would be an elevated expression of metallothionein, a cysteine rich protein that can bind up to 7 divalent metal ions, in low-grade prostatic tumours while high grade tumours were negatively stained for this protein (17), as well as the inhibition of the selenoenzyme thioredoxin reductase in cancer cell lines treated with chemopreventive or anticancer drugs (18, 19). For these reasons here we studied the metalloproteome of prostate tissue obtained from healthy rats and that of the human prostate carcinoma cell line DU-145. To consider whether this is an appropriate model or not, the total proteomes were compared primarily by two-dimensional electrophoresis (2-DE). Generally, the spot pattern varied significantly between the cell line and the rat prostate gland, although it was possible to match some larger spots (triangles in Figure 1a and 1b). This variance might be a result from either the different sample species; the higher oxidative stress present in cell culture or the cancerous properties of the cell line.

An increased risk of developing leukaemia or lymphoma (20), lung (21) and prostate cancer (22, 23) has been correlated with a low selenium status and hence prostate from selenium-deficient rats was used as additional samples with minor proteomic variances (Figure 1c).

By its presence in selenoproteins such as glutathione peroxidase, selenium-functions as an antioxidant and therefore the influence of selenium deficiency on the total antioxidant capacity was examined. As can be seen in Figure 1d, there was no significant difference between the selenium-adequate and the selenium-lacking prostate, while in the cancer cells, grown under hyperoxic conditions, the value was elevated nearly two-fold.

From these results, selenium deficiency was considered to be an adequate example for further investigations provided that the observed results are consistent with those observed for the cell line. SEC-ICP/MS is a widely favoured tool for the investigation of metal-biomolecule complexes in biological samples (24, 25), e.g. protein-bound trace elements in different organs were studied (26) and elevated metallothionein-bound cadmium concentrations in urine from bladder carcinoma patients were detected by this method recently (27).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

2-DE spot patterns from: a, selenium-adequate rat prostate; b, DU-145 cell line; and c, selenium-deficient rat prostate, were compared. Triangles indicate spots matched between a) and b). In d) the total antioxidative capacity expressed as mmol ascorbic acid equivalents per g protein for the three kinds of sample have been determined.

SEC-ICP/MS was applied to cytosolic prostate fractions from selenium-adequate and selenium-deficient fed rats and DU-145 cells. In this way, a number of peaks were detected for the majority of monitored elements. Even so, a minority of chromatographs were reproducible between individuals of the same type of sample. Chromatographs with low variance can be found in Figure 2 and will be described here in brief only: For both nickel and zinc, two peaks were detected which are suppressed in the carcinoma cell line and in selenium deficiency. In copper spectra, three peaks with retention times between 13 and 37 minutes were observed for all samples. An additional intense signal, with a retention time of 10 minutes, was present in the cell line. No chromium peak was expressed in the carcinoma sample, while three peaks with differing intensities were noted in the rat samples. Furthermore, for cobalt an increase in a low molecular weight fraction (retention time: 25 minutes) in the cells and in selenium deficiency was accompanied by a decreasing high molecular weight peak at about 15 minutes.

Finally, it was demonstrated by western blotting and immunochemistry that copper and zinc binding superoxide dismutase (Cu/Zn-SOD) as well as metallothionein were up-regulated in DU-145 as compared with the rat samples, while signal for thioredoxin reductase was most intense in selenium adequate prostate (Figure 3).

Discussion

A number of metal- or metalloid-binding proteins have been identified as being regulated by the presence of tumours (17, 27, 28). However, only few studies refer to alterations in the metalloproteome of cancer samples. Therefore, by means of SEC-ICP/MS and immunochemical tests, prostate tissue and prostate adenocarcinoma cells were investigated. Their significant proteomic differences, demonstrated by 2D-E, motivated us to introduce selenium deficient prostate as an additional sample offering a spot pattern very similar to the control animals, while the correlations that were found earlier between a low selenium status and the development of cancer (20-23) suggest that there may exist some proteomic changes towards the cancerous state. To ensure that the observed variations are not a result of elevated oxidative stress due to the absence of selenoproteins, we determined the total antioxidative capacity in the samples. Similar values were detected between selenium-adequate and selenium-deficient animals. In contrast, the nearly two-fold increase in DU-145 cells could be explained by growth of the cell culture under hyperoxic conditions as compared to physiological oxygen pressure (29, 30). This elevated pro-oxidant level also provides explanation for the intense signal present at about 17.0 minutes in the copper chromatograph, which was suspected most likely to be dimeric SOD.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Size-exclusion chromatography with ICP/MS detection. The three chromatographs in each line represent from left to right: selenium-adequate prostate; selenium-deficient prostate and DU-145 cells. The element ratios from top to bottom are: Ni60/Rh103; Zn66/Rh103; Cu63/Rh103; Cr52/Rh103 and Co59/Rh103, where Rh103 was an internal standard.

The finding of that peak recognised by separating a mixture of purified Cu/Zn-SOD together with rat control samples substantiated our hypothesis (data not shown). Moreover, by immunochemistry, SOD was detectable in DU-145 only. The maximum at about 10 minutes present in the human cell line might also be Cu/Zn-SOD since it was reported that a single amino acid exchange between human (Val42) and rat (Asp48) SOD promotes dimerisation of the latter instead of tetramerisation of the first (31).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Immunochemical detection of Cu/Zn-SOD (a), thioredoxin reductase (b) and metallothionein (c) in rat prostate and DU-145 cells.

There is apparent discrepancy between the limited number of peaks detected for zinc and the wide distribution of zinc-binding proteins. In fact, the biological functions of this element arise from hundreds of enzymes and thousands of zinc finger protein domains (32), a large population of which bind to nucleic acids and function as transcription factors and hence are not expected to be present in the cytosolic fraction investigated. On the contrary, the predominant species present at 17.5 minutes may overlap less abundantly. Earlier observations (33) detected a diminished concentration of zinc in prostate cancer but not in healthy tissue; in addition, we demonstrated that in both states, the vast majority of the element is bound to proteins. However, the decrease of zinc-binding proteins is not accompanied by a suppressed expression of metallothionein which was even induced in DU-145 cells. Combining this data and the results obtained in (27), it could be speculated that the elevated metallothionein-bound cadmium would be caused by simultaneous lack of zinc supply and induction of the protein.

Finally, in our investigations cobalt was bound to a higher percentage to a low molecular weight (LMW) fraction in cancer cells and selenium-deficient prostate. Since accumulation of vitamin B12 was observed in tumours (34), we related the LMW signal to cobalamin. The higher molecular weight peak (about 25 kDa) was preliminarily assigned to the cytoplasmic homologue of the Nudix16 hydrolase; NUDT16L1. In the presence of Mn²⁺ or Co²⁺, Nudix16 removes CAP structures from a broad range of RNA substrates and with Mg²⁺ only U8 snoRNA is accepted (35). 5.8S rRNA, a down-stream product of U8 snoRNA, is covalently linked to p53 and possibly an important target in the control of ribosome function during cell differentiation and oncogenesis (36). Because in the report of Hultdin et al. (37) it is also mentioned that elevated plasma vitamin B12 levels are accompanied by an enhanced prostate cancer risk, the redistribution of cobalt may point to an unknown mechanism of cell survival in carcinogenesis.

From our obtained results we concluded that although metalloproteome possesses high variance for many elements – maybe as a result of varying trace element status-some element profiles offered characteristic changes in cancer cells and selenium-deficient animals. This includes a decrease of nickel and zinc peaks, in particular, as well as the disappearance of a cobalt binding species of about 25 kDa and the appearance of a low molecular weight fraction (presumably cobalamin).