In the present study, 101 PC patients who were histopathologically diagnosed after surgical resection at the Department of Gastrointestinal Surgery and Pathology (Sun Yat-sen University Cancer Center) from January 1999 to December 2010 were retrospectively enrolled. Gemcitabine-based chemotherapy was administered to the 44 patients with advanced-stage disease after surgery; none of the patients received radiotherapy. The clinicopathological features of the patient cohort are listed in Table I. All patients were followed-up on regular basis; last follow-up was May 2011 with a mean follow-up time of 19 months (range, 1–122 months), during which time 84 cancer-related deaths occurred. Four fresh PC and paired adjacent non-cancerous pancreatic tissues were collected for real-time quantitative polymerase chain reaction (qRT-PCR) and western blot analysis. All of the patients provided consent for the use of their paraffin embedded tissues for research purposes.

In order to use these clinical materials for research-purposes, the patient’s prior consent and approval from the Institutional Research Ethics Committee of the Sun Yat-sen University Cancer Center were obtained. Clinicopathological classification and staging were determined according to the criteria proposed by the American Joint Committee on Cancer and International Union Against Cancer criteria.

The human immortalized pancreatic ductal epithelial cell line (IPEC) hTERT-HPNE E6/E7 and PC cell lines, including BxPc-3, Capan-2, SW1990, CFPAC-1 and PANC-1 were obtained from the American Type Culture Collection (Manassas, VA, USA). All cell lines were maintained in DMEM medium (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT, USA).

For depletion of PPARγ, two human short hairpin RNA (shRNA) sequences were individually cloned into the pSuper- retro-neo plasmid to generate pSuper-retro-PPARγ-RNAi(s), respectively; the sequences were RNAi#1: GCGGAGATCTCC AGTGATATC and RNAi#2: GCTGAATGTGAAGCCCAT TGA (synthesized by Invitrogen). Retroviral production and infection were performed as previously described (28). Stable cell lines expression PPARγ or PPARγ short hairpin RNAs (shRNA) were selected for 10 days with 0.5 μg/ml puromycin.

Total RNA was extracted from cultured cells using TRIzol reagent (Invitrogen) following the manufacturer’s instructions, and cDNA was amplified and quantified by quantitative real-time PCR (qRT-PCR) using the ABI PRISM 7500 Sequence Detection System (Applied Biosystems, Grand Island, NY, USA) with SYBR Green I (Molecular Probes, Grand Island, NY, USA). The qRT-PCR primers were selected as follows: PPARγ, forward 5′-GAGTACCAAAGTGC AATCAAAGTG-3′ and reverse 5′-TCTCCACAGACACGAC ATTC-3′. Expression data were normalized to the geometric mean of house-keeping gene GAPDH (forward: 5′-ACCACA GTCCATGCCATCAC-3′ and reverse: 5′-TCCACCACCCTGT TGCTGTA-3′) to control the variability in expression levels and calculated as, where C_t represents the threshold cycle for each transcript.

Immunohistochemical analysis was performed to evaluate PPARγ protein expression in the 101 human fine-needle aspirations of PC tissues and 4 fresh pancreatic tissues. In brief, paraffin-embedded specimens were cut into 4-μm sections, baked at 60°C for 2 h, deparaffinized with xylene, rehydrated, subjected to antigen retrieval by microwaving in EDTA antigen retrieval buffer, and treated with 3% hydrogen peroxide in methanol to quench endogenous peroxidase activity, followed by 1% bovine serum albumin to block non-specific binding. The sections were incubated with rabbit anti-PPARγ (1:100; Santa Cruz Biotechnology, Santa Cruz, CA, USA) overnight at 4°C. Normal goat serum was used as a negative control. After washing, the tissue sections were incubated with biotinylated anti-rabbit secondary antibody (Zymed, San Francisco, CA, USA) followed by strep-tavidin-horseradish peroxidase complex (Zymed). The sites of immunoreactivity were visualized using the 3.3′-diami-nobenzidine (ZSGB-Bio, Beijing, China) and the sections were counterstained with 10% Mayer’s hematoxylin (Sigma-Aldrich, St. Louis, MO, USA), dehydrated and mounted.

PPARγ staining was scored as: i) the percentage of positive tumor cells in the tumor tissue (0, 1–5%; 1, 6–25%; 2, 26–50%; 3, 51–75%; 4, 76–100%) and ii) the staining intensity: (0, no signal; 1, weak; 2, moderate; 3, strong). The immunoreactivity score (IRS) was calculated by multiplying the score for the percentage of positive cells and the intensity score (possible range, 0–7) (29). IRS scores ≥3 were considered high expression. IHC staining was quantitatively analyzed using the Axio Vision Rel.4.6 computerized image analysis system assisted with an automatic measurement program (Carl Zeiss, Oberkochen, Baden-Württemberg, Germany). Briefly, the stained sections were evaluated at ×200 magnification, and 10 representative stained fields in each section were analyzed to determine the mean optical density (MOD), which represents the strength of the staining signal as the percentage of positive pixels. The MOD data were analyzed using the t-test to compare the average MOD difference between different groups of tissues; P<0.05 was considered statistically significant.

Cells were incubated at 37°C in an incubator with 5% CO₂. Cells at 0.2×10⁴ of each line were seeded in a 96-well microtiter plate and allowed to adhere to the plate for 24 h at 37°C. Cells were then treated for 72 h at 37°C with either gemcitabine (Eli Lilly and Co., Indianapolis, IN, USA) or 5-fluorouracil (5-FU) (Sigma-Aldrich) with or without Pioglitazone (Pio, Takeda Pharmaceutical Co. Ltd., Osaka, Japan) at various concentrations, as indicated in the figure legends. At the appropriate time-points, the cells were incubated with 100 μl 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) dye (0.5 mg/ml; Sigma-Aldrich) for 4 h, the culture medium was removed and 150 μl of dimethyl sulfoxide (DMSO) (Sigma-Aldrich) was added to dissolve the formazan crystals. The absorbance values were measured by Spectra Max M5 spectrometer (Molecular Devices, Sunnyvale, CA, USA) at 570 nm, with 630 nm as the reference wavelength. All experiments were carried out in triplicate. In MTT assay, there is a linear relationship between the OD reading and the number of viable cells. Percent drug killing of cancer cells = (OD_{control well} - OD_{drug-exposed well}) / (OD_{control well} - OD_{blank well}) × 100%. The average OD readings were obtained from 3 duplicate wells in any one MTT assay, all experiments were carried out in triplicate.

Early and late cell apoptosis was measured by flow cytometry using Annexin V and PI provided in a commercial kit (Biovision, Zurich, Switzerland) according to the manufacturer’s protocol. Briefly, the cells were seeded in a 6-well plate at a density of 1×10⁶ cells/well and remained in spontaneous culture media at 37°C with 5% CO₂ for 24 h at 37°C. Then the cells were treated for 48 h at 37°C with either gemcitabine or 5-fluorouracil with or without Pio at various concentrations as indicated in the figure legends. On the test day, 1×10⁵ trypsinized cells were washed twice in PBS and resuspended in 100 μl of binding buffer, then suspended in Annexin V-binding buffer, stained with Annexin V-FITC for 15 min at room temperature, washed and stained with PI. The samples were analyzed using a FACSCalibur flow cytometer equipped with CellQuest-Pro software (Becton-Dickinson, San Jose, CA, USA).

The fresh tissues were ground to a powder in liquid nitrogen, lysed with sampling buffer [62.5 mmol/l Tris-HCl (pH 6.8), 2% SDS, 10% glycerol and 5% 2-β-mercaptoethanol], and the protein concentrations were determined using the Bradford assay (Bio-Rad Laboratories, Berkeley, CA, USA). Equal amounts of protein were electrophoretically separated on 9% polyacrylamide SDS gels (SDS-PAGE) and transferred to polyvinylidene fluoride membranes (Amersham Pharmacia Biotech, QC, Canada). The membranes were incubated with an anti-PPARγ mouse antibody (1:100; Santa Cruz Biotechnology), followed by a horseradish peroxidase-conjugated anti-mouse IgG antibody (1:2,000; Amersham Pharmacia Biotech) and the bands were detected using an enhanced chemiluminescence kit (Amersham Pharmacia Biotech) according to the manufacturer’s instructions. The membranes were subsequently stripped and re-probed with an anti-tubulin mouse monoclonal antibody (1:2,000; Sigma-Aldrich) as a loading control.

All statistical analyses were carried out using SPSS version 13.0 (SPSS, Chicago, IL, USA). Comparisons between groups were performed using two-tailed paired Student’s t-tests. The relationships between PPARγ expression and the patient clinicopathological features were analyzed using the χ² test. Survival curves were plotted by the Kaplan-Meier method and compared using the log-rank test. Survival data were evaluated using univariate and multivariate Cox regression analyses. P-values <0.05 were considered statistically significant for all analyses.

The association between PPARγ expression level and clinicopatholigical characteristics in PC patients was studied. PPARγ expression was examined in 101 fine-needle aspirations of PC tissues, including 21 cases of clinical stage I (20.8%), 53 cases of stage II (52.5%), 15 cases of stage III (14.9%) and 12 cases of stage IV (11.9%) PC. Significant correlation was observed between PPARγ expression level and several prognostic risk factors such as clinical stage, T classification and N classification (Table I). PPARγ immunostaining was localized in the nucleus. The IHC analysis demonstrated that PPARγ was markedly upregulated in PC tissues, but it was only marginally detectable or not at all in normal pancreatic tissues (P<0.05, Fig. 1A). Quantitative analysis of the IHC staining using the MOD scores indicated that PPARγ expression in clinical stage I to stage IV primary PC was significantly higher than normal pancreatic tissues (P<0.05, Fig. 1B).

The PC patients were divided into two groups: high and low PPARγ expression based on the IRS score determined during IHC analysis (high = IRS score ≥4). Kaplan-Meier and log-rank analysis demonstrated a significant difference in the overall survival time of patients with low and high PPARγ expression. High PPARγ expression was closely associated with poor overall survival (P<0.001, Fig. 2A). Furthermore, high PPARγ expression correlated strongly with poor overall survival in the subsets of patients with T3+T4 disease or N0 disease (both P<0.001; Fig. 2B and C). Interestingly, high PPARγ protein expression also correlated significantly with poorer overall survival in patients who received postoperative treatment for advanced disease (P<0.001, Fig. 2D). PPARγ protein expression correlated significantly with the clinical stage, T classification, N classification, overall survival and survival of patients with chemotherapy regimen after surgery (Table II). Moreover, univariate and multivariate analyses revealed that clinical stage, the T classifications and PPARγ expression were each recognized as independent prognostic factors (Table III), suggesting that PPARγ expression can be utilized as a predictor of survival in PC patients, and also that patients with low levels of PPARγ expression might experience greater benefit from adjuvant therapy.

To investigate the potential role of PPARγ in the progression of PC, Western blotting, qRT-PCR and IHC analyses were performed on PC cell lines (BxPc-3, Capan-2, SW 1990, CFPAC-1, PANC-1), immortalized pancreatic epithelial cells (IPECs) and four freshly isolated PC tissues and the paired adjacent non-cancerous tissues. As shown in Fig. 3A and B both PPARγ mRNA and protein were markedly upregulated in all of the PC cell lines tested, compared with IPECs. The expression of PPARγ mRNA was 4- to 10-fold higher in PC tissues than the adjacent non-tumor tissues (Fig. 3C). Furthermore, comparative analysis revealed that PPARγ protein expression was upregulated in all four of the freshly isolated PC tissues, compared with the matched adjacent non-tumor tissues (Fig. 3D) suggesting that PPARγ is overexpressed in PC.

To investigate the impact of PPARγ on the efficacy of chemotherapy in PC, we used PPARγ ligand pioglitazone and two short hairpin RNAs for PPARγ was employed to suppress endogenous PPARγ expression stably in BxPc-3 and PANC-1 cell lines. Western blot analysis revealed that the amount of PPARγ protein in PPARγ RNAi(s) cells, normalized by α-tubulin, was reduced up to 80% compared with PPARγ RNAi-vector cells (Fig. 4A). The PANC-1 and BxPc-3 cells were exposed to Gem alone or Gem plus PPARγ ligand (10 μM Pio) with or without PPARγ knockdown. After 72-h treatment, greater apoptosis was observed in the cells treated with Gem plus PPARγ ligand, compared with the cells treated with Gem alone. The PPARγ-RNAi(s) cells treated with Gem plus PPARγ ligand showed minimal apoptosis, compared with the vehicle treated with Gem alone or Gem plus PPARγ ligand. Similar results were observed in the cells exposed to 5-FU alone or 5-FU plus PPARγ ligand (10 μM Pio) with or without PPARγ knockdown (Fig. 4B). As shown in Fig. 4C, silencing of endogenous PPARγ increased Gem plus Pio IC₅₀ values 3.5- and 1.7-fold for Panc-1 and BxPc-3 cells, increased 5-FU plus Pio IC₅₀ values 2.6- and 2.4-fold, respectively. Futhermore, the PPARγ-RNAi(s) cells Gem or 5-FU plus Pio IC₅₀ values significantly lower than the vector cells incubated with Gem or 5-FU alone (P<0.01). These results indicate that downregulation of PPARγ decreased the cytotoxic effects of Gem and 5-FU in pancreatic cancer cells.

As shown in Fig. 5A–D, downregulated of PPARγ inhibited the percentage of apoptotic PANC-1 cells after the cells were treated in 500 nm Gem plus 10 μM Pio for 48 h, compared with the PPARγ-RNAi vector cells treated with Gem alone or Gem plus Pio, respective (P<0.01). Similarly, the apoptosis rate of the PANC-1 PPARγ-RNAi(s) cells treated with 5-FU (40 μM) plus Pio (10 μM) clearly lower than the PPARγ-vector cells treated with 5-FU plus Pio or with 5-FU alone, which were (3.7±0.3, 4.0±0.5), (33.0±1.8) and (18.2±2.2)%, respectively (P<0.01, Fig. 5E–H). Similar situation appeared when BxPc-3 PPARγ-RNAi(s) cells and PPARγ-RNAi vector cells were treatment with drugs: the precentage of early and late apoptotic BxPc-3 PPARγ-RNAi(s) cells treated with Gem/5-FU plus Pio were clearly lower than the PPARγ-vector cells, also lower than the PPARγ-vector cells treated with Gem/5-FU alone (P<0.01, Fig. 6). These results indicate that downregulation of PPARγ decreased the antitumor effects on pancreatic cancer cells and is a potent apoptosis inhibitor.