Purine Nucleoside Phosphorylase mediated molecular chemotherapy and conventional chemotherapy: A tangible union against chemoresistant cancer

Docetaxel (Commercial name: Taxotere) (Aventis, Pharmaceuticals Inc, NJ), Carboplatin (Pfizer, NSW, Australia) and Fludarabine Phosphate (Fludara), (Schering-Plough, Germany) were used. Antibodies to Bcl-2, Bik, Bax, Survivin, Caspase-7/9 and PARP were used (source, Additional file 1(Table S1).

Ovarian cancer cell lines were maintained either in Dulbecco's Modified Eagle Medium (DMEM) or Roswell Park Memorial Institute medium-1640 (RPMI) media supplemented with 10% foetal calf serum (FCS), 50 U/mL Penicillin and 50 μg/mL Streptomycin (GIBCO/Invitrogen, VIC, Australia) at 37°C, 5% CO₂, in a humidified incubator.

Replication defective Ad vectors expressing either Green Fluorescent Protein (AdGFP) or PNP gene (AdPNP) under the control of cytomegalovirus (CMV) promoter were constructed (PNP or GFP genes are cloned in the E1 region of Ad5 genome), propagated and titrated in accordance with the instructions from the AdEasy™ adenoviral vector kit (Stratagene, TX, USA).

Cells (24 well plates) infected with Ad-vectors at different multiplicities of infection (moi, plaque forming units (pfu)/cell) were assessed for gene expression at 48 h post infection (pi). GFP: For AdGFP infected cells, the GFP expression was determined by flow-cytometry using CellQuest™ (Version 3.0) software (Becton Dickinson, San Jose, CA). PNP: That the toxic effects of the PNP gene are only possible in the presence of Fludara has been shown unequivocally (with appropriate controls) in our previous studies and thus, is an acceptable measure of PNP activity [11, 26]. PNP activity, as determined by the enzymic assay, is accompanied by viral dose dependent ability of the prodrug, Fludara, to kill the transduced cells. The conversion of Fludara to its toxic metabolites by PNP in infected cells was shown by high performance liquid chromatography. Hence in this study, expression of PNP (same as used in our previous studies) in AdPNP infected cells was determined by assessment of cell killing in the presence of the prodrug, Fludara, given 48 h post infection using cell viability assays; after 3 to 7 days of incubation with the prodrug a colorimetric assay using the REDOX (4-[3-(4-Iodophenyl)-2-(4-nitrophenyl)-2H-5-tetrazolio]-1, 3-benzene Disulfonate (WST-1) dye (Takara Pty Ltd. Otsu Shiga, Japan) was used. At relevant times, cells were incubated in media containing WST-1 (10: 1) for 2 h and the absorbance measured at 450 nm (Tecan Sunrise, Phoenix Research Products, USA).

Cells treated with different concentrations of docetaxel and/or carboplatin were assessed for cell viability using WST-1 assay at different times (see 2.4).

For synergy experiments, cells (plated in triplicate in 96 well plates, also see Additional file 2 (information, A1) were infected with either AdPNP or AdGFP. After 48 h, virus-containing media were replaced with Fludara and/or docetaxel and/or carboplatin containing media. The un-infected cells were treated with docetaxel and/or carboplatin. After 3 -5 days, cell viabilities were assessed by WST-1 assay.

Treated or untreated cells were analysed for apoptosis using the M30 CytoDEATH™assay kit (Additional file 1(Table S1) for source), based on binding of an antibody to a Caspase-cleaved epitope of cytokeratin 18-cytoskelatal protein in apoptotic cells and not in viable or necrotic cells (Peviva AB, Bromma, Sweden). Briefly, cells (1 × 10⁵) (24 well plate) given PNP-GDEPT and/or docetaxel and/or carboplatin for 48 h were fixed in methanol (-20°C for 30 minutes), washed twice (PBS with 0.1% Tween 20 (PBST) and incubated with 100 μL of M30 CytoDEATH™ (1:100 dilution) or isotype control IgG_2b (1:125) antibodies in incubation buffer (PBS with1% bovine serum albumin and 0.1% Tween 20). After a wash in PBST, cells were incubated for 1 h (4°C) with 100 μL of the fluorescein-labelled secondary antibody (FITC 1:70) (Silenus, Melbourne, Australia) and then analysed by flow cytometry (FlowJo Version 7.2.2 (Tree Star, Inc., CA).

Subconfluent OVCAR-3 cell cultures treated with different treatments were harvested at the indicated time points (0.025% EDTA), washed in ice-cold PBS and fixed using ice-cold ethanol at 4°C for 30 minutes. After two washes in PBS, cellular DNA was stained with propidium iodide solution (50 μg/mL propidium iodide, 0.1 mg/mL RNase A and 0.25% Tween 20 in PBS) for 1 h at 37°C. The percentage of cells in the G₀/G₁, S and G₂/M phases was assessed using flow cytometry.

Proteins in cell lysates (50 μg) from variably treated cells were resolved on 10% polyacrylamide gel by SDS-PAGE and then blotted on to nitrocellulose membranes [30]. The blots were incubated overnight in the relevant primary antibody at concentration recommended by the manufacturer (source, Additional file 1 (Table S1) followed by 1 h incubation with secondary antibody (anti-rabbit IgG Horse Radish Peroxidase (HRP) conjugated, 1:1000 or anti-mouse IgG HRP, 1:5000 (source, Additional file 1 (Table S1). The proteins were detected using an enhanced chemiluminescence (ECL) kit (Piercenet, Il, USA); treated membranes were exposed to X-ray film from 1-60 min (as required)-and developed using standard protocols. The protein bands were quantified from these films by densitometry using Quantity One software (Bio-Rad, Hercules, CA).

Untreated OVCAR-3 cells and those treated with PNP-GDEPT (AdPNP at the moi of 10 pfu/cell for 48 hours and then incubation with Fludara at 1 μg/mL (2.7 μM) were lysed (3 cycles of freeze thawing) and cell lysates (50 μg protein) were resolved by SDS-PAGE using 10% polyacrylamide gel. After Coomassie blue staining (G-250 stain, Bio-Rad, NSW, Australia), from each lane ~ 10-20 gel pieces were excised and peptides extracted (trypsin digest) were processed as described [31] (Mass Spectrometry Unit, UNSW, Sydney, Australia). The digested peptides were separated using HPLC (Ultimate 3000 HPLC and autosampler system, Amsterdam, Netherlands), concentrated and desalted and then subjected to mass spectrometry (MS) (LTQ FT Ultra (Thermo Electron, Bremen, Germany).

Before evaluating the AdPNP-GDEPT, the permissivity of ovarian cancer cell lines to Ad transductions was evaluated by assessing the efficiency of Ad/CMV/GFP transductions in different cell types at 48 h post infection. This showed a variable level of Ad5-uptake (Figure 1A). At the moi of 100 pfu/cell, OVCAR-3 cells were the most permissive (% GFP expressing cells ± SEM: 80 ± 5) followed by A-2780 (15 ± 4) and SKOV-3 cells (12 ± 4). A highly permissive, lung cancer cell line, A-549, was used as a positive control. Overall, OVCAR-3 cells displayed the highest Ad-permissivity at all Ad/CMV/GFP-doses tested, whilst, Caov-3 cells were almost Ad-refractory with only 3% GFP expressing cells at the high moi of 500 pfu/cell.

Evaluation of AdPNP-GDEPT (Fludara: 1 μg/mL (~2.7 μM) in infected cells showed a dose dependent increase in cytotoxicity which correlated with the Ad-permissivity of cell lines (Figure 1B). The toxic effects (% cell death compared to untreated controls ± SEM) were maximal in OVCAR-3 cells (50% ± 5 at moi of 4 pfu/cell) followed by SKOV-3 (25% ± 5 at moi of 100 pfu/cell) and A-2780 (15% ± 7 at moi of 100 pfu/cell) with no effect in Ad-refractory Caov-3 cells. This was also reflected in the IC₅₀ values (± SEM) [calculated as moi of AdPNP for 50% cytotoxicity]. Interestingly, the 'bystander effects' also correlated with the Ad-permissivity of cells; in highly permissive OVCAR-3 cells, a significant level of cell inhibition was observed at low viral doses of 1 or 2 moi and up to 90-100% cell growth inhibition was noted when only 12% (at moi of 10, Figure 1A) of cells expressed PNP. Whilst this differed from a previous report [10] that as low as 1% infection was required to kill all cells, the differences may be due to the expression system used, and the use of different cell lines.

An insight into the molecular changes underlying the cytotoxic effects of PNP-GDEPT may help to understand its interactions with other modalities. Shot-gun proteomics analyses were performed on proteins resolved by SDS-PAGE using cell lysates from untreated and treated (AdPNP at moi of 10 pfu/cell plus 2.7 μM Fludara) platinum-resistant OVCAR-3 cells (see section 2.10). From these, a list of up- and down-regulated proteins was generated (Tables 1 & 2); (Note: for description and potential roles of these proteins, see Additional file 4 (Tables S3 and S4). Normalisation of data showed symmetrical distribution, suggesting the similarity of the relative abundance of most proteins in both samples; however, clearly distinctive data points representing potentially unique or differentially abundant proteins were obvious. The stringency of analyses and the low level variation between the two data sets suggested the reliability of these data. Ratio values of treated to untreated cells (range 7925-26529, P < 0.01) indicated clear upregulation of 16 proteins, to levels significantly higher than those predicted for 99% confidence (232.9632 ± 163.636). The abundant expression of E. coli PNP, only in PNP-GDEPT treated samples (Scores: 26239 [treated] vs. none [untreated]) provided evidence of its production in Ad/CMV/PNP infected samples. Although, modulated expression of several proteins was seen, only those present in only one sample and which may have a potential role in cancer progression or apoptosis, are listed.

Overall, PNP-GDEPT treatment lead to general down-regulation of proteins involved in (1) cellular metabolism (lipid, amino acid, carbohydrate and glycolysis) (e.g. polymerase (RNA) II (DNA directed), dihydrolipoamide S-succinyltransferase (E2 component of 2-oxo-glutarate complex), ribosomal protein L4, mitochondrial trifunctional protein), (2) oncogenesis or cancer progression (cadherin, desmoplakin, plakoglobin, karyopherin, spondin, agrin, GTP binding protein and cadherin 6) and (3) drug resistance (antiquitin and epoxide hydrolase). An upregulation of proteins involved in apoptosis, tumour suppression Poly (ADP-ribose) polymerase (PARP) and dead box polypeptide 3) and general DNA synthesis was noted.

Growth inhibitory effects of docetaxel against all four ovarian cancer cells lines at different times were variable and dose and time dependent (Figures 2A, and Additional files 5 and 6 (Figure S1 & Table S5). Overall, A-2780 cells were the most sensitive followed by SKOV-3-, OVCAR-3- and Caov-3 cells. SKOV-3 and OVCAR-3 cells were chosen for the following experiments based on their mid-range sensitivity to docetaxel, variable Ad-permissivity and platinum resistance. This allowed us to test the synergies in different scenarios in cisplatin resistant ovarian cancer cells, which would be especially relevant for patients resistant to platinum treatment. The IC₅₀ values for docetaxel (Figure 2A) showed SKOV-3 cells (0.31 nM ± 0.7; R²:0.96) (Figure 2) to be more docetaxel-sensitive compared to the OVCAR-3 cells (0.61 nM ± 1.1; R²:0.99).

Use of cisplatin resistant lines would be particularly relevant to studies involving carboplatin, given its efficacy against cisplatin resistant ovarian cancer [32]. Using a range commonly used in other studies (IC₅₀ range ~5 μM-35 μM [33], carboplatin efficacy was found to be dose dependent (Figure 2B). However, only a slight difference in the carboplatin sensitivity (IC₅₀) was noted (OVCAR-3: 43.12 μM ± 6; R²:0.97 and SKOV-3: 38.73 μM ± 4; R²: 0.96) (Figure 2B).

Cell viability (WST-1-based) evaluations of the cells treated with the combinations including docetaxel plus carboplatin, docetaxel plus PNP-GDEPT, carboplatin plus PNP-GDEPT and combination of all three, in general, showed a significant, dose dependent inhibition of cell growth (Additional file 7 (Figure A2) An assessment of interactions between modalities through CalcuSyn analysis [27, 28] (see methods for details) of cell viability data generated Combination Index-Fraction affected (CI-Fa) plots. These data showed that at the combination ratio (based on potency) of 1:1, the combined efficacy was additive for the docetaxel/carboplatin combination (Figure 3A and Table 3) (CI:0.89 at IC₅₀ dose; R value 0.99), synergistic when PNP-GDEPT was combined with either docetaxel (Figure 3B and Table 3) (CI: 0.21 at IC₅₀; R value 0.93) or carboplatin (Figure 3C and Table 3) (CI:0.28 at IC₅₀; R value 0.96) and strongly synergistic when all three were combined (Figure 3D and Table 3) (CI:0.11 at IC₅₀; R value 0.94). Statistical analyses further revealed the "Dose reduction Index (DRI)" values for each modality in combination to achieve a specific effect for the two cell lines (Additional file 8 (Table S6). In comparison to the individual doses, a dose reduction of up to ~20 fold for PNP-GDEPT, ~27 fold for docetaxel and ~31 fold for carboplatin was predicted to generate 50% cell killing when used in trimodal combination. In addition to evaluation of combinations at the ratio of 1:1, effects were also evaluated at other ratios e.g. 1:2, 2:1, 1:4 (Additional file 9 (Table S7, CI values). Overall, the trends were similar to those obtained when drugs were combined at 1:1 ratio with some variations. To assess whether the drug interactions lead to long term efficacy, clonogenic assays were performed comparing the effects of trimodal therapy with single treatments. Given the fact that carboplatin+docetaxel is the standard treatment for ovarian cancer patients, long term efficacy of the trimodal combination would have direct relevance to the clinic. Treatment of OVCAR-3 cells with, docetaxel (0.6 nM) or carboplatin (40 μM) or PNP-GDEPT (AdPNP at moi of 4 pfu/cell plus 2.7 μM Fludara) alone reduced the number of colonies to ~42%A, 36% and 22% respectively, compared with untreated controls (Figure 4A), whilst the combination of all three reduced colony growth to 2.3%+/-1.3% (13+/-7 cells vs. 560) (Figure 4A). This is slightly greater than an additive effect which would have yielded 3% (0.42 × 0.36 × 0.22 = 0.03) of the controls (17+/-13 cells compared to 560). Within the experimental errors of a clonogenic assay, where it is difficult to accurately assess clonal numbers especially in the untreated cells and due to other factors such as low plating efficiency, colony size and cancer heterogeneity, these results indicate an improvement over additive effects. Thus, though the trends are clear, statistical significance could not be achieved and the final confirmation of the potential of these combinations can only be drawn from exhaustive in vivo evaluations. While, these experiments were beyond the scope of this specific study, in another study, we have shown that our in vitro findings comparing synergistic responses of prostate cancer cells to PNP-GDEPT with docetaxel (including long term effects as evaluated by clonogenic assays) were supported by the in vivo outcomes (Singh et al, Clin Cancer Res. 2011, In press).

For these assessments, time points beyond 48 h were not included as then the significant cellular toxicity (apoptosis/cell death) observed with combination regimens could potentially obscure the molecular responses to different treatments [34]. Evaluation of apoptosis in SKOV-3 cells treated with docetaxel (1.5 nM), carboplatin (20 μM) and PNP-GDEPT (AdPNP moi: 150 pfu/cell, 2.7 μMFludara) (doses showing best efficacy) either alone or in combination showed that the numbers of apoptotic cells increased in a time dependent manner in tri-combination treated cells (Figure 4B). The percentage of M30 positive cells (a marker of early apoptosis) was maximal in tri-combination treated cells (49 ± 7) followed by bi-modal (docetaxel/carboplatin (17 ± 5), docetaxel/GDEPT (23 ± 6), carboplatin/GDEPT (31 ± 5), and then individual treatments (PNP-GDEPT (11 ± 4), docetaxel (14 ± 4), carboplatin (16 ± 6). As anticipated, apoptosis achieved in Fludara only treated cells was low (1.5 ± 1), In contrast, when the Fludara was converted by PNP as in cells treated with the combinations including PNP-GDEPT, there was a relatively higher number of apoptotic cells irrespective of the duration of the treatment.

Cell cycle analyses of OVCAR-3 cells treated for 48 h with docetaxel (0.6 nM), carboplatin (10 μM) and PNP-GDEPT (AdPNP moi: 10 pfu/cell, 1 μg/mL(2.7 μM) Fludara) either alone or in combination showed a decline in percentage of cells in G1 phase (20-35%) relative to controls, irrespective of the treatments applied (Figure 4C). As expected [35, 36] carboplatin (G2/M: 33% ± 9) and docetaxel (G2/M: 43% ± 8) treatments led to an increase in G2/M populations. The percentage of apoptotic cells (sub-G1phase) for different treatments was maximal in tri-combination treated cells (37% ± 6) followed by bi-modal (carboplatin/docetaxel (28% ± 6) < PNP-GDEPT/docetaxel (32% ± 6) < PNP-GDEPT/carboplatin (34% ± 7) and then least in single agent treated cells (carboplatin (9% ± 3) < docetaxel (12% ± 6) < PNP-GDEPT (20% ± 5) with negligible apoptosis in control treated cells. As with M30 analysis, the percentage apoptosis was greater when treatments included PNP-GDEPT. Results were statistically significant when combination treatment was compared with single treatment or control treatment (P ≤ 0.05).

Mutations in pathways involved in apoptosis modulate their responses to therapy (e.g. chemoresistance) [reviewed in [37]. Apoptosis involves initiation (apoptotic stimulus, e.g., chemotherapeutic or biological agent), effector and execution phases (which decide the fate of the cell) [38]. Thus, relative expressions of protein members representing effector phase [Bcl-2, Bik, Bax, Inhibitor of apoptosis (survivin) and execution phase (caspase-7/9, PARP) were evaluated in the treated cells (Figure 5A & 5B) (See Additional file 10 (Table S8) for their correlation with ovarian cancer outcomes). A significant but variable down-regulation of anti-apoptotic genes/proteins (Bcl-2) and up-regulation of pro-apoptotic genes (Bax, Bik and Bok) was achieved when modalities were combined, especially when PNP-GDEPT was included (Figure 5, Additional file 11 (Table S9). Their expression was relatively little affected by either carboplatin or docetaxel used alone or together. Survivin was down-regulated by all combination treatments. Evaluation of PARP (cleaved and un-cleaved) expression in differentially treated cells showed the upregulation of un-cleaved (especially 116 kDa band) and cleaved PARP in response to all treatments; this was even more in the case with combination treatments, and highest when PNP-GDEPT was involved. As Caspase-mediated proteolysis of PARP is a biochemical marker of apoptosis that denotes the final stages of apoptosis leading to DNA fragmentation [39], this indicates that cell killing was occurring in the treated cells.