Phenotype-dependent apoptosis signalling in mesothelioma cells after selenite exposure

This study was carried out using a well-established model system for mesothelioma differentiation, consisting of the two cell sub-lines STAV-AB and STAV-FCS. Cells were derived from a single tumor, and subsequently induced to differentiate into the epithelioid (STAV-AB) and the sarcomatoid phenotype (STAV-FCS), respectively, by altering the serum composition [30]. Hence, STAV-AB cells were grown in Gibco RPMI 1640 medium (Invitrogen) and 10% human AB serum, whereas STAV-FCS cells were grown in the same medium and 10% fetal calf serum. The specific differentiation of these cells has been evidenced by immunoprofiling showing that STAV-AB cells express more cytokeratin, whereas STAV-FCS cells have stronger reactivity to vimentin antibodies [21] as well as by morphometry. The elongated sarcomatoid cell morphology of the STAV-FCS cells and the more round epithelioid morphology of the STAV-AB cells have been confirmed by average length:width ratios of 3.42 in the STAV-FCS cells and 1.58 in the STAV-AB cells [31]. Jurkat cells were obtained from the American Type Culture Collection (ATCC) and grown in RPMI 1640 medium and 20% fetal calf serum. All cells were grown at 37°C with 5% CO₂ and passaged approximately twice per week.

To investigate the contributions of several signalling pathways, inhibitors were used against key enzymes. Cells were washed, trypsinized and re-seeded with the respective inhibitors (specified in table 1) 24 h prior to selenite treatment, except for the JNK-inhibitor, with which they were pre-incubated for 1 h. Selenite was then added to the medium and the cells were harvested 24 or 48 h later. Titrations were performed with all inhibitors based on the manufacturers' instructions and concentrations reported in the literature. In all cases, the highest non-toxic doses tested were accepted.

Viability assays were performed in conjunction with flow cytometry experiments to obtain internal controls. Aliquots of cell suspensions prepared for flow cytometry were plated in triplicates in 96-well plates, with a density of approx. 5000 cells per well. They were then analysed using the WST-1 assay (Roche), whereby a tetrazolium salt is cleaved by mitochondrial enzymes to yield a coloured product, to measure viability. The plates were read in a Spectramax spectrophotometer at 450 nm with subtraction of background absorbance at 600 nm.

Flow cytometric assays for detection of apoptosis were carried out using the Annexin V kit (Caltag Laboratories) according to the manufacturer's protocol. Briefly, trypsinized cells were resuspended in Binding Buffer with Annexin V-FITC and Propidium Iodide (PI), and incubated for 15 minutes in the dark.

For analysis of Bax expression, cells were fixed in 0.25% paraformaldehyde, and permeabilised with 100 μg/ml Digitonin. Aliquots were then incubated for 30 minutes with phycoerythrin-conjugated mouse anti-human Bax antibody (Santa Cruz, sc20067) or mouse IgG as a control, both in final dilution 1:10. Morphological controls were established using cytospins. Slides were fixed in 4% buffered formaline, washed 2 × 5 min in PBS, and air-dried. Staining was performed with the same antibody concentration and incubation time, and the staining was evaluated by confocal microscopy.

For analysis of Bcl-XL expression, cells were fixed in 1% formaldehyde, and permeabilised with 0.1% Saponin. Aliquots were then incubated for 15 minutes with phycoerythrin-conjugated rabbit anti-human Bcl-XL antibody (GeneTex, GTX46035), final dilution 1:800, or rabbit IgG as a control. The secondary antibody Alexa 488 goat anti-rabbit IgG was diluted 1:1600, and incubation was performed for 15 minutes in the dark.

The mitochondrial membrane potential was measured using two independent methods. 1) The Mitochondria Staining Kit (Sigma) was used according to the manufacturer's instructions. Briefly, cells were trypsinised and then resuspended in a solution of 45% medium, 5% serum and 50% staining solution containing the JC-1 probe. They were incubated for 20 min in 37°C, and then washed with staining buffer. Cells treated with Valinomycin were used as a positive control. 2) With the fluorescent probe DiOC6(3) (3,3-dihexyloxacarbocyanine iodide; Molecular Probes), cells were incubated for 15 minutes with concentrations ranging from 1 to 100 nM DiOC6(3). After staining, an aliquot of cells was prepared for confocal microscopy to verify that the staining was localized to the mitochondria.

For analysis of procaspase-3 expression, cells were fixed in 1% paraformaldehyde, and permeabilised with 10 μg/ml Digitonin. Aliquots were then incubated for 30 minutes with a rabbit monoclonal antibody to procaspase-3 (Abcam, ab32150), final dilution 1:150, or rabbit IgG as a control. The secondary antibody Goat anti-Rabbit IgG-FITC (Abcam, ab6717) was diluted 1:300, and incubation was performed for 30 minutes in the dark.

Detection of the active form of caspase-3 was performed with a FITC-conjugated antibody (BD Biosciences, 559341). Cells were fixed in 1% paraformaldehyde, and resuspended in 100 μg/ml Digitonin solution with antibody in final dilution 3:20, and incubated for 30 minutes at 4°C. Cells treated with 2 μM doxorubicin for 24 h were used as positive controls.

Flow cytometry was always performed immediately after the staining was completed. All analyses were performed on a Becton Dickinson flow cytometer and the data were processed in the Cell Quest program.

The M-30 Apoptosense assay (Peviva, Stockholm) measures cytokeratin 18 fragments specifically produced by cleavage by caspases 3, 6, 7, and 9 [32, 33]. Cells were seeded in 96-well microtiter plates with or without 10 μM selenite and 0.2 μg/ml doxorubicin. After 24 h, cells were lysed by the addition of 10 μl 10% Tergitol-type NP-40 (Sigma-Aldrich) to each well. The ELISA analysis was carried out according to the manufacturer's instructions. Briefly, 25 μl samples were incubated together with 75 μl horseradish peroxidase-conjugate solution on the ELISA microplate for 4 h on a shaker. 200 μl of tetramethylbenzidine substrate solution were added and the plate was incubated for a further 20 min. The reaction was stopped by the addition of 50 μl 1.0 M H₂SO₄, and the absorbance at 450 nm was determined on a Spectramax spectrophotometer.

For analysis of nuclear translocation of p53 and p21, cytospins were prepared. For p53 analysis, the slides were fixed in ice-cold dry acetone. Prior to staining, they were heated to 100°C for 5 min in citrate buffer, pH 6.0. Staining was performed using the p53 Refine kit (Novacastra). For p21 analysis, the slides were fixed in 4% buffered formaline, and air-dried. Staining was performed with a mouse monoclonal antibody (Calbiochem, OP64), diluted 1:200, for 30 minutes.

For analysis with monodansyl cadaverine (MDC), cells were grown on sterilised Superfrost Plus slides (Menzel GmbH &Co). The slides were stained for 10 minutes with 10 μM MDC (BioChemica), and immediately analysed by confocal microscopy.

Nuclear extracts were prepared as described previously [34]. Electrophoretic Mobility Shift Assay (EMSA) was conducted using the LightShift Chemiluminescent EMSA Kit (Pierce). 20 μg of nuclear protein was used for each sample. The double-stranded oligonucleotide probes for the p53 binding site (sense 5'-TACAGAACATGTCTAAGCATGCTGGGG-3') were annealed and labeled with biotin. To label DNA probes, the Biotin 3' End DNA Labeling Kit (Pierce) was used according to the manufacturer's protocol.

ELISA was used to quantify the amounts of thioredoxin (Trx) in the cells. The assay was adapted from Pekkari et al [35]. Wells were coated with a primary monoclonal antibody (2G11, kindly provided by dr. Anders Rosén of the University of Linköping) overnight at 4°C, 5 μg/ml diluted in carbonate buffer, pH 9.6. Secondary biotinylated antibody (IMCO Corporation) was added in a concentration of 5 μg/ml. Absorbance at 405 nm was measured using a SpectraMax 250 spectrophotometer (Molecular Devices). Data were analyzed using the SOFTmax Pro software, v. 2.6.

All experiments were performed at least three times. When one representative experiment is shown, it was chosen on the basis of being closest to the average of all the experiments performed. Student's t-test, two-way ANOVA with Dunnett's post test or Bonferroni's multiple comparison test, and χ²-tests were used to determine statistical significance. The choice of statistical method for each experiment is indicated in the respective legend text.

More than 90% of the untreated cells were viable, appearing in the lower left quadrant. Representative Annexin-PI plots are shown in figures 1A and 1B. Baseline early apoptosis (cells in the lower right quadrant), averaged over three experiments, was 4% in the epithelioid cells (Figure 1C) and 9% in the sarcomatoid cells (Figure 1D). Note that the figures 1A and 1B show only one representative experiment. 10 μM selenite decreased the viable fraction particularly in the sarcomatoid cells, as has also been reported previously [1]. This finding was confirmed by viability assays (data not shown). Selenite also increased the proportion of early apoptotic cells (Figures 1C and 1D). There were around 15% of cells in the upper quadrants in both cell types after selenite treatment, representing cells late in their apoptotic process or undergoing necrosis. A time-course experiment with measurements up to 48 h was performed to verify that 24 h was a suitable time-point for analysis (data not shown).

Loss of mitochondrial membrane potential (δΦ_m) is associated with apoptosis. Following selenite treatment, a 3-fold loss of δΦ_m was seen in both cell types (Table 2). These findings were confirmed by DiOC₆(3) staining, and the specificity for mitochondria was verified using confocal microscopy (data not shown). The loss of δΦ_m in both phenotypes after selenite treatment agrees well with earlier studies [15, 19, 36].

To further delineate the role of signalling molecules among the MAP kinases and cathepsins, chemical inhibitors against these enzymes were used (Table 1). In the untreated epithelioid cells, the inhibitors decreased the baseline apoptotic fraction by 20–50% [see Additional file 1]. This demonstrates the efficacy of the inhibitors at the concentrations in which they were used. None of the enzyme inhibitors affected the proportion of viable cells during Annexin-PI apoptosis assays, although the WST-1 viability assays indicated a modest growth inhibitory effect of CA 074-Me and SB 203580 (data not shown).

Further controls to verify the efficacy of the chemical inhibitors were obtained by testing them on Jurkat cells over a 25 h time course following apoptosis induction with 0,2 μM staurosporine. The inhibitors of JNK, p53 and cathepsin D and E successfully decreased the apoptosis induction, whereas the cathepsin B inhibitor increased it [see Additional file 2].

In the sarcomatoid cells, the p38 inhibitor SB203580 caused a small decrease in the apoptotic response to selenite (Figure 1D). In the epithelioid cells, p38 inhibition had no effect on the ability of selenite to induce apoptosis. However, selenite caused an even more marked drop of the δΦ_m after p38 inhibition in both cell types (Table 2). This indicates that p38 was involved in apoptotic signalling particularly in the more sensitive sarcomatoid cells. The effect of inhibition was small however, and it cannot be regarded a key pathway. Activation of p38 after selenite exposure has previously been shown in cervix [18], leukemia [42] and prostate cancer cells [5].

Inhibition of JNK increased the proportion of selenite-induced early apoptotic cells by more than two thirds in the epithelioid cells (Figure 1C). In the sarcomatoid cells the effect was comparable to that without the inhibitor (Figure 1D). Scant effect on the loss of δΦ_m was observed (Table 2). JNK apparently played no role in apoptosis signalling in the sarcomatoid cells. In the epithelioid cells, JNK even had a small antiapoptotic effect. The lack of proapoptotic activity is concordant with earlier findings in cervix cancer cells [18] but different from findings in prostate cancer cells [5].

Immunocytochemistry revealed that both epithelioid and sarcomatoid cells responded to selenite with a time-dependent increase of nuclear p53 immunoreactivity. After 24 h, the proportion of positive cells was increased approximately 1.5-fold (Figure 2A–E), and after 48 h, approximately 2-fold (not shown). EMSA analysis showed, however, that p53 exhibited less binding to DNA after selenite treatment (Figure 3B). Thus, although selenite caused nuclear accumulation of p53, it also decreased the DNA-binding activity. This result was surprising, as p53 has been implicated as a mediator of selenite-induced apoptosis signalling in other cell systems [5, 17, 18, 43, 44].

The expression of p21 (also known as Cip1 and WAF1) in response to genotoxic stress is tightly regulated by p53 (reviewed in [45]), and we therefore measured it as an additional indicator of p53 activity. The fraction of p21-positive cells was approximately doubled by selenite treatment (Figure 2F–J). Although these changes are statistically significant, the positive fraction was very small even after selenite treatment. As a positive control, epithelioid cells were treated with 2 μM doxorubicin and showed a 22% positive fraction (not shown).

Cells of either phenotype treated with the p53 inhibitor Pifithrin did not show a decreased apoptosis frequency as judged by Annexin-PI (Figure 1), nor a smaller loss of δΦ_m (Table 2). This is particularly interesting since p53 inhibition decreased the baseline apoptosis in untreated cells (Figure 1, Additional file 1). Consequently, p53 was active in the control cells but was inactivated by selenite. Apoptosis was still induced by selenite, implicating p53-independent pathways in this process.

To find the mechanism of inhibition, we considered the complex regulation of p53 activity. The central DNA-binding core domain of p53 contains one zinc atom. Zinc chelators have been shown to cause accumulation of wild-type p53 in a structurally aberrant form with inhibited DNA-binding activity [46]. Selenium is a known chelator of zinc and when applied in vivo as selenite or its reduced form selenide, it forms nanocrystals of zinc-selenium with free or loosely bound zinc [47]. Another possibility is that selenite as an oxidizing agent may act directly upon regulatory cysteines on the p53 molecule, leading to an accumulation of oxidized p53 incapable of DNA-binding [48]. Also, secondary mediated redox regulation needs to be considered. The multifunctional protein Redox Effector Factor 1 (Ref-1) is involved in the redox regulation of stress inducible transcription factors such as Activating Protein-1, Nuclear Factor-κB, Hypoxia Inducible Factor-1 and p53, and may play an important role in this system. Ref-1 depends on thioredoxin (Trx) to maintain its active reduced state [49‐51]. In a yeast experimental system, it has been shown that deletion of thioredoxin reductase (TrxR) downregulates p53 activity by keeping it in its oxidized form [52, 53]. Trx overexpression on the other hand has been shown to increase p53 transactivation of reporter genes in human cell lines [49].

Protein levels of Trx were reduced by selenite treatment in sarcomatoid cells, from 175 ng/mg to 100 ng/mg. The epithelioid cells had a baseline expression of 57 ng/mg, decreasing slightly to 52 ng/mg after selenite treatment (Figure 3). It has been demonstrated previously that levels of TrxR decrease significantly in both cell lines with selenite treatment [1]. Consequently, redox inactivation of p53 is a plausible explanation for the lack of activity that was seen despite nuclear accumulation following selenite exposure.

The immunoreactivity for the proapoptotic mediator Bax increased significantly in the sarcomatoid cells but not in the epithelioid cells following selenite treatment (Figure 4). This clear phenotypic difference may partially explain why sarcomatoid cells are more sensitive to selenite. Morphological controls verified that staining was localised to cytoplasmic granules consistent with mitochondria (not shown). Although activation of Bax in response to selenite has been shown in other systems [9, 17, 18, 44, 54], this is the first report of differential expression coupled to sensitivity.

In mesothelioma, the antiapoptotic Bcl-2 family member Bcl-XL is frequently overexpressed [55], and this has been shown to be an important mechanism by which mesothelioma cells gain apoptosis resistance [56]. In the epithelioid cells, the Bcl-XL expression decreased markedly after selenite treatment, whereas only a subpopulation of the sarcomatoid cells showed lower expression after treatment (Figure 4).

Both epithelioid and sarcomatoid cells showed a 6-fold increase in caspase-mediated cleavage of cytokeratin 18 after selenite treatment (Figure 5), indicating activation of caspases 3, 6, 7, and 9. Doxorubicin, as a positive control, caused a 10-fold increase in the epithelioid cells and a 6-fold increase in the sarcomatoid cells.

Flow cytometric analysis for procaspase-3 showed that both cell types have a similar baseline expression. After selenite treatment, subpopulations of both phenotypes lose procaspase-3. In the epithelioid cells, this corresponds to the appearance of a distinct subpopulation (13%) that is positive for activated caspase-3. In the sarcomatoid cells, there is also a small fraction (5%) of cells that stain more intensely for activated caspase-3, but it is not distinctly separated from the main peak (Figure 6). Caspase-3 activation was confirmed in doxorubicin treated-controls: 49% of epithelioid cells were positive, and 20% of sarcomatoid cells (not shown).

Divergent data have been published regarding the role of caspases in selenite-induced apoptosis. Several studies have shown that selenite causes a caspase-independent apoptotic cell death [6, 18, 40], whereas others have shown caspase-dependence [9, 17, 36, 57]. We report that caspase-3 was activated in a sub-population of epithelioid cells, but little reactivity was seen in sarcomatoid cells. The limited caspase activation in sarcomatoid cells was surprising. A possible explanation could be an upregulation of Inhibitor of Apoptosis (IAP) family members such as survivin and XIAP. Earlier studies have found that overexpression of IAP family members is common in mesothelioma cells [58‐61].

Cathepsins are a group of proteases that are physiologically present in lysosomes, and may be released upon stimuli such as oxidative stress [62]. Cells that were pretreated with cathepsin B inhibitor CA-074 Me showed slightly less apoptosis after selenite exposure (Figure 1). In the sarcomatoid cells, this was reflected in correspondingly increased viability. In the epithelioid cells, the viable proportion decreased slightly instead. Interestingly, when selenite was combined with the cathepsin B inhibitor, the loss of δΦ_m was greater than with any other inhibitor (Table 2). Cathepsin D and E inhibitor Pepstatin A did not affect the induction of apoptosis by selenite, nor did it alter the loss of δΦ_m.

Autophagy is a form of programmed cell death in which cells do not exhibit apoptotic characteristics. Kim et al have shown that selenite induces autophagy in glioma cells [38]. We wanted to investigate whether some of the cell death that we observe could be due to autophagy. Cells were stained with monodansyl cadaverine and analysed with confocal microscopy for the appearance of granules that might represent autophagic vesicles. No changes in the staining pattern could however be detected between untreated and selenite-treated cells of either phenotype [see Additional file 3].