Pioglitazone was purchased from Interchim (Montluçon, France). ΔPGJ2 was from Cayman Chemicals (Bertin, Montigny le Bretonneux, France). 2’-7’-dichlorofluoresceine diacetate (DCF-DA) was provided by Accros Organics (Halluin, France), acridine orange, carbonyl cyanide m-chlorophenylhydrazone (CCCP), propidium iodide, GSH, 4,5-diaminofluorescein diacetate (DAF2-DA), orthophtalaldehyde (OPA), L-buthionine sulfoximine (BSO) and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) were from Sigma-Aldrich (Saint-Quentin Fallavier France). Rabbit antibodies against caspase-3 (1087-1), GCL (5529-1), HO-1 (1922-1), NQO1 (S2173), PARP (1078-1), Beclin-1 (2026-1), lamin A/C (3770-1) were purchased from Epitomics (Euromedex, Souffelweyersheim, France). Mouse monoclonal antibodies against GAPDH (MAB374) were purchased from Millipore (Mosheim, France). Human GSTPi (F-6), human Nrf2 (H-300) and human Trx (which recognize Trx1 and 2 isoforms; 1H6H6) antibodies were from Santa Cruz Biotechnologies (Perray en Yvelines, France). Mouse or rabbit HRP-conjugated second antibodies were purchased from Santa Cruz Technologies. RNAse A (from bovine pancreas) was from Euromedex. Other chemicals were of analytical grade.

Δ2-pioglitazone was synthesized according to the procedure of Sohda et al (11), except for the first step of condensation which was described by Deguest et al (12).

HT29 and HCT116 cells were grown in Dulbecco’s minimum essential medium (DMEM, Eurobio, Courtaboeuf, France) supplemented with 10% (v/v) heat-inactivated (30 min at 56°C) fetal calf serum (Eurobio), 50 μg/ml gentamycin (Invitrogen, Paisley, UK) and 2 mM L-glutamine (Eurobio). Cells were maintained at 37°C in a humidified atmosphere in the presence of 5% CO₂. Seeding was 10⁵ cells/ml in all experiments, with the exception of the MTT procedure. Medium was changed daily 48 h after seeding.

ΔPio cytotoxicity was assessed by the MTT procedure. Cells were seeded at 10⁴ cells/well in 96-well plates and treated with increasing concentrations of ΔPio (0–100 μM) for 72 h. The stock solution of ΔPio was 50 mM in DMSO. Dilutions of the drug were performed in DMSO prior to addition in the cell medium. Control cells were treated with 0.1% (v/v) of DMSO, used as a molecule diluent. MTT (0.5 mg/ml) was prepared in medium containing fetal calf serum, and 100 μl/well was added, and each plate was incubated for 2 h at 37°C. Formazan precipitate was dissolved in DMSO and absorbance at 540 nm was read.

Cell growth kinetics with ΔPio (50 μM) or Pio (50 μM) were established for HT29 and HCT116 cells. Cells were seeded in 6-well plates and treated for 4 days. Control cells were treated with 0.1% (v/v) of DMSO. Then, cells were harvested by trypsination [0.02% (w/v) trypsin/2 mM EDTA solution, Eurobio], and 100 μl of cell suspension was mixed to an equal volume of 0.04% (v/v) trypan blue solution prepared with Dulbecco’s phosphate-buffered saline (DPBS, Eurobio) and living cells were counted using a Malassez hematometer. Cell growth was also tested with ΔPio or Pio together with 1 mM ascorbic acid for three days. Cells were harvested and counted. Results are expressed as the ratio of treated cells versus control cells cultivated with vehicle (mitotic index).

HT29 or HCT116 cells were seeded in 6-well plates. Forty-eight hours after seeding, cells were exposed to each drug for 24 h. Cell layers were washed twice with DPBS, harvested by trypsination and fixed in 70% (v/v) ethanol solution for 2 h and stored at −20°C. Cells were then centrifuged at 1,000 g for 5 min at 4°C, washed with DPBS and centrifuged again. Pellets were suspended in 500 μl DPBS containing 50 μg/ml propidium iodide, 20 μg/ml of RNAse A and 0.1% (v/v) Triton X-100 for 20 min and analysed by FACS (FL2A; FACSCalibur, BD Sciences, Le Pont de Claix, France). Results were quantified using the CellQuest software (BD Sciences) and cell cycle distribution was analysed using Modfit software (Verity Software House, Topsham, ME, USA).

Anchorage-dependent clonogenic assay was performed in 6-well plates. Cells were seeded at 1,000 cells/well in complete medium. ΔPio or Pio were added to cell suspensions after seeding. Cells were left to grow for 7 days. Cell clones were then washed twice with DPBS and fixed in 70% (v/v) cold-ethanol for 15 min. They were washed twice with DPBS before adding 1% (v/v) toluidine blue solution diluted in DPBS for 15 min and washed with DPBS. Images were captured with GelDoc (Bio-Rad, Marne la Coquette, France).

Two days after seeding, cells were treated with ΔPio or Pio for various times up to 24 h. Cells were loaded with 50 μM DCF-DA for 15 min or 1 μM DAF-2DA for 30 min at 37°C. After incubation, cells were harvested by trypsination, washed with DPBS and finally suspended in 1 ml of the same buffer. Fluorescence was measured in 30,000 cells/sample by flow cytometry with excitation and emission settings, at 488 and 530/30 nm (FL1). In addition, 5 mM N-acetyl cysteine or 1 mM ascorbic acid, as antioxidant, was added to the medium 1 h before drug treatment. Production of reactive species from mitochondria was assessed also in the presence or absence of the protonophore carbonyl cyanide m-chlorophenylhydrazone (CCCP). Cells were treated with each drug and 50 mM CCCP was added for 15 min at 37°C. Cells were loaded with 50 μM DCF-DA for 15 min and they were prepared for FACS analysis. Data from control (DMSO-treated cells) and drug-treated cells were compared. Results are expressed relative to those obtained from control cells taken as 100.

Intracellular GSH levels were determined by an HPLC method described by Lenton et al (13) with slight modifications using a C18-Thermo Hypersil column (Fisher Scientific, Illkirch, France) and 1.5 mM orthophtalaldehyde. Cells were grown in 25 cm²-flasks. Briefly, 48 h after seeding, cells were exposed to 50 μM Pio, 50 μM ΔPio or 3 μM ΔPGJ2 for various times, up to 24 h. When used, antioxidant was added 1 h before drug treatment. Cell layers were washed three times with ice-cold DPBS, then suspended with 10% (v/v) perchloric acid ice-cold solution/2 mM EDTA, pH 8.0. The homogenates were centrifuged at 15000 g for 15 min at 4°C. The supernatant were stored at −80°C until HPLC analysis. Perchloric acid-precipitated proteins (pellets) were solubilized in daily prepared 1 M NaOH prior to protein content determination. GSH contents were calculated as nano-mole of GSH/mg of protein. Data from DMSO-treated cells (control cells) and treated cells were compared and they are expressed relative to those obtained from control cells, taken as 100.

Cells were seeded in 25 cm²-flasks. Total protein homogenates were prepared from drug- and DMSO-treated cells as followed: cell layers were washed twice with ice-cold DPBS and they were scrapped with 25 mM HEPES/KOH, pH 7.5 containing 10 mM EDTA, 400 mM KCl, 0.5% (v/v) Igepal, 1 mM DTT and 0.1% (v/v) of a protease inhibitor mixture (Sigma). Homogenates were collected and left on ice for 30 min before centrifugation at 15,000 g for 15 min at 4°C. Supernatants were stored at −80°C until use. Nuclear and cytoplasmic protein extracts were obtained from drug and DMSO-treated cells grown in 90-mm dishes. Cell layers were washed twice in cold DPBS and scrapped gently in DPBS. The suspension obtained was centrifuged at 1,000 g for 2 min at 4°C. The cell pellet was mixed with an equal volume of 10 mM HEPES/KOH, pH 7.5 containing 60 mM KCl, 1 mM DTT, 0.5% (v/v) Igepal and 0.1% (v/v) of a protease inhibitor mixture, and left on ice for 5 min. The sample was centrifuged at 1,000 g for 5 min at 4°C. The supernatant, which correspond to the cytoplasmic fraction, was further centrifuged for 15 min at 15,000 g and at 4°C and the pellet discarded. The cytoplasmic fractions were stored at −80°C. The pellet obtained after the initial centrifugation was suspended with HEPES/KOH buffer without detergent and centrifuged at 8,000 g for 2 min at 4°C. This step was repeated at least three times. After washing, each pellet was suspended in 25 mM Tris-HCl, pH 8.0 containing 600 mM KCl, 1 mM DTT and 0.1% (v/v) protease inhibitor mixture (Sigma). They were left on ice for 15 min, vortexed every 5 min and then centrifuged at 15,000 g for 15 min at 4°C. The supernatant corresponded to the nuclear fraction and each sample was stored at −80°C until use.

Protein (25 μg) from whole cell homogenates or 20 μg of protein from nuclear or cytoplasmic fractions were resolved in 10–15% SDS-PAGE and transferred onto PVDF membranes. The saturation step was performed in 50 mM Tris-HCl, pH 7.4 containing 0.15 M NaCl, 5% (w/v) non-fat milk and 0.01% (v/v) Tween-20. The antibody solutions were prepared in the same buffer and incubated overnight at 4°C by gentle agitation. Washing steps (3 times for 5 min) were performed with 50 mM Tris-HCl, pH 7.4 containing 0.15 M NaCl and 0.01% (v/v) Tween-20. The blots were then incubated in this buffer containing the diluted second antibody (horseradish peroxidase conjugated to goat anti-rabbit (1:7,500) or anti-mouse (1:7,500), Santa Cruz Technologies), and finally washed 3 times for 5 min with the same buffer. Blots were developed by chemiluminescence detection according to the manufacturer’s protocol (Santa Cruz Technologies).

HT29 and HCT116 cells were seeded in 6-well plates and treated as described above for 1–3 days with daily medium changes. In parallel, HT29 cells were treated with 500 μM of hydrogen peroxide used as an inducer of autophagy. Cells were incubated with 1 μg/ml acridine orange for 15 min, then harvested by trypsination and suspended in DPBS before FACS analysis. Fluorescence was measured in 30,000 cells/sample, with excitation at 488 nm and emission settings at 670 nm (FL3). Results were analysed using CellQuest and Cyflogic software and they were compared to those obtained with DMSO-treated cells, taken as 1.0. In parallel, cells were treated with each drug in the presence or absence of 5 mM NAC over three days. Total protein homogenates were prepared and western blots were performed as described, using Beclin-1 (1:1,000) and GAPDH (1:10,000) antibodies, respectively, as a marker of autophago-some formation and an internal marker for protein loading.

Differences between results from control (DMSO-treated cells) and treated cells were analysed by Student’s t-test and any difference was considered significant at P<0.05.

In a first set of experiments, we assessed whether ΔPio was efficient in inhibiting colon cancer-derived cell growth since the molecule had little effect on prostate carcinoma cells (10). ΔPio cytotoxicity was estimated by the MTT procedure and the trypan blue coloration method. The IC₅₀ was, respectively, 53.7±2.8 and 46.2±2.4 μM in HT29 and HCT116 cells. We established cell growth kinetics for HT29 and HCT116 cells treated with 50 μM ΔPio over four days; the results were compared to those obtained after cell exposure to 50 μM Pio (Fig. 2A). HT29 cells in the presence of ΔPio reached the plateau phase of growth after three days of treatment. In contrast, cell growth was decreased up to 35% at the end of the experiment when HT29 cells were treated in the presence of Pio (Fig. 2A). A 40%-decrease was obtained when HCT116 cells were exposed to each molecule, but the effect of ΔPio, previously found in HT29 cells, was not observed with these cells (Fig. 2B). Anchorage-dependent assays were performed with both cell lines (Fig. 2C) and the results obtained confirmed the higher anti-proliferative property of ΔPio. The impact of ΔPio treatment on cell growth was associated with the alteration of cell distribution in cell cycle phases. Cell exposure to this drug resulted in cell accumulation in G0/G1-phase, whereas Pio treatment enhanced cell distribution in S-phase (Table I). However, HT29 or HCT116 cell growth arrests were not associated with apoptosis cell death. As shown in Fig. 3, neither caspase-3 activation nor PARP-1 cleavage was found in protein homogenates prepared from ΔPio or Pio-treated cells. In contrast, protein cleavage was obtained when HT29 cells were treated for two days in the presence of 20 μM camptothecin used as an inducer of apoptosis.

Several lines of evidence demonstrated that cell exposure to TZD generated oxidative stress which dramatically impacted cell survival (14). We tested whether ΔPio treatment involved the production of ROS and RNS in colorectal cells. Fluorescent probes, i.e. DCF-DA and DAF-2DA for ROS and RNS detection, respectively, were used to quantify any changes in the levels of reactive species after HT29 cell exposure to ΔPio over 24 h. ROS levels reached a peak 15 min after drug treatment (Fig. 4A). ROS levels were markedly increased with the maximum reached at 8 h and it decreased slightly over 24 h (Fig. 4B). Similarly, we evaluated ROS level after cell exposure to Pio. In Pio-treated cells, ROS level stayed constant over two hours, then decreased to a level lower than ROS content in DMSO-treated cells (Fig. 4B). When HCT116 cells were tested, ROS levels were increased after a 10-min drug treatment (Fig. 4C), but they stayed mostly constant over the experimental duration whatever the molecule tested (Fig. 4D). Since mitochondria are the main source of radical species and notably superoxide anion radical, colorectal cells were treated in the presence of CCCP together with ΔPio and Pio then loaded with DCF-DA (Table II). CCCP pre-treatment inhibited mostly Pio or ΔPio-mediated production of ROS in HCT116 cells and to a lesser extent in HT29 cells, supporting the finding that part of the redox alteration was associated with the mitochondria. Moreover, pre-treatment with ascorbic acid or NAC (as antioxidants) prevented or limited the formation of reactive species in colorectal cells (Table II). When cells were treated over three days with ΔPio or Pio together with 1 mM ascorbic acid, cell growth was partially restored, suggesting that cell growth arrest was associated to the production of excessive ROS (Fig. 5A and B, respectively). We used ascorbic acid instead of NAC since we observed that 5 mM NAC treatment inhibited cell growth as reported by others (15). Using DAF-2DA, we demonstrated that RNS were present within HT29 cells. The production reached a maximum after 6 h and stayed constant thereafter (Fig. 6). Of note, the effect of ΔPio on RNS production was greater than Pio treatment in HT29. RNS production was increased in HCT116 cells exposed to each molecule after 2 h, but RNS levels stayed almost constant over the experiment duration (Fig. 6B). The results obtained were associated with the increase of iNOS level in both cell lines (Fig. 6C). In fact, HCT116 cells expressed iNOS constitutively whereas the expression of iNOS is inducible in HT29 cells. Nevertheless, our results suggested that RNS were produced in our cell models. The production of an excess of reactive species has been associated to the depletion of intracellular GSH content (14). As demonstrated in Fig. 7, GSH level was decreased in a time-dependent manner over 24 h. They were estimated at 30–50% when colorectal cells were treated with ΔPio or Pio for 24 h (Fig. 7A and 7B), respectively.

Among the mechanisms related to the generation of oxidative stress and involved in cell growth arrest or survival, we studied whether ΔPio treatment could induce autophagy. The formation of acidic vesicles was analysed by acridine orange red-fluorescence detection by FACS (Fig. 8A and B). The increase of red-fluorescence due to the acidic vesicles was time-dependent and ΔPio treatment had a greater effect as compared to cells exposed to Pio. Beclin-1, a marker of autophagosome formation, was induced after cell exposure to ΔPio or Pio for 3 days whatever the cell line tested (Fig. 8C and D). Treatment with ΔPio had a more pronounced effect when compared to cells exposed to Pio. NAC pre-treatment lowered Beclin-1 expression confirming that autophagy was associated to drug-mediated oxidative stress in both cell lines (Fig. 8E and F). However, NAC treatment did not rescue completely the effect of ΔPio over three days.

Activation of the Nrf2/Keap1 pathway is associated with redox changes within cells (7). We studied whether ΔPio treatment activated translocation of Nrf2 in the cell nuclei. These results were compared to those obtained after cell exposure to ΔPGJ2, known as a potent inducer of the Nrf2/Keap1 pathway (16) and the endogenous ligand of PPARγ (17). However, the molecule has been shown to induce apoptosis at a concentration higher than 10 μM in colorectal cells (18) but also in cells from other tissue origins (19). We assessed whether cell exposure to 3 μM ΔPGJ2 had any impact on colorectal cell growth and we showed that ΔPGJ2 treatment did not trigger cells to enter apoptosis (Fig. 3). As demonstrated by western blots performed with nuclear extracts, 3 μM ΔPGJ2 treatment enhanced dramatically Nrf2 nuclear translocation in both cell lines throughout the experiment (Fig. 9). ΔPGJ2 treatment enhanced HO-1, a Nrf2 target gene, at the protein level (Fig. 10C and F, respectively). Maximum HO-1 levels were achieved when HCT116 cells are exposed to ΔPGJ2 (25 times induction) for 24 h, as compared to that found in HT29 cells (4.7 times induction). Similarly, NQO1 was detected in HT29 and HCT16 cells and the protein levels were increased after cell exposure to ΔPGJ2. Moreover, GCL protein contents were also enhanced in ΔPGJ2-treated cells and it was associated with the increase of intracellular GSH levels starting 8 h after cell treatment (Table III). Addition of BSO, a specific inhibitor of GCL activity, abrogated ΔPGJ2-mediated GSH synthesis. Thus, at the dose used ΔPGJ2 had a protective effect on colorectal cells. As shown in Fig. 9, Nrf2 accumulated in cell nuclei from ΔPio or Pio-treated cells but differences existed depending on the cell line tested. In HT29 cells, maximum Nrf2 nuclear levels were found at 1 h in ΔPio-treated cells and at 2 h in Pio-treated cells (Fig. 9A). In contrast, Nrf2 translocation began 1 h after the HCT116 cell exposure to each drug and it stayed mostly constant during the experiment duration (Fig. 9B). We analysed which Nrf2 target gene could be activated during drug treatment. Whatever the cell line tested, HO-1 expression was induced demonstrating that HO-1 was involved in ΔPio-mediated oxidative stress. ΔPio or Pio-treatment increased HO-1 expression to a similar level whatever the cell tested (Fig. 10A,B and D,E respectively). Focusing on known Nrf2 target genes such as NQO1, GSTPi, thioredoxin and GCL, respective protein levels depended on the cell line tested and the treatment applied. NQO1 expression was increased at the protein levels, in Pio (Fig. 10A,D) and ΔPio treated cells (Fig. 10B,E). There were only small changes in GCL levels in Pio and ΔPio-treated HT29 or HCT116 cells (Fig. 10A,B and D,E respectively). GSTPi expression did not change dramatically upon molecule treatment whatever the cell line tested. Trx levels rose to a maximum after HCT116 cell exposure to Pio for 4 h (Fig. 10D), whereas its expression was often decreased in the other treatment conditions. Thus, activation of the Nrf2 pathway correlated well to an increase of reactive species production within the cells.

HO-1 expression is enhanced in stressed cells but its expression depends also on activation of diverse transduction pathways or a mechanism involving GSH depletion (7). Based on our results, notably the presence of reactive species and GSH depletion during ΔPio treatment, we assessed whether HO-1 expression could be modulated by antioxidants such as NAC (a thiol scavenger) (Fig. 11). Pre-treatment with NAC prevented the production of ROS within HT29 colorectal cells (Table II) and addition of the antioxidant inhibited almost completely ΔPio-induced HO-1 expression (Fig. 11A). These results were close to those obtained when HCT116 cells were tested (Table II and Fig. 11B).