Activation of Peroxisome Proliferator-activated Receptor γ by Troglitazone Inhibits Cell Growth through the Increase of p27^Kip1 in Human Pancreatic Carcinoma Cells¹

Pancreatic adenocarcinoma is one of the most lethal malignancies and now ranks fourth and fifth as a cause of cancer death in men and women, respectively, in the United States (1). A majority of pancreatic cancer patients present with metastatic disease or advanced local disease, precluding a curative surgical resection. Chemotherapy has not resulted in a significant survival benefit, with a median survival of 4.1 months (2). Based on these facts, a new molecular target is needed for treatment of pancreatic carcinomas.

PPARγ,³a nuclear hormone receptor, provides a strong link between lipid metabolism and the regulation of gene transcription(3, 4, 5). Recent studies show that PPARγ is expressed at high levels in human colonic mucosa and colon carcinoma cells(6, 7, 8). Ligand activation of PPARγ in human colon carcinoma cells causes growth inhibition (6, 7, 8). In contrast, other groups have demonstrated that activation of PPARγpromotes the development of colon tumors in mice (9, 10). Thus, the role of PPARγ activation in the cellular behavior of colon tumors is controversial. With regard to other tumors, PPARγactivation induced growth arrest in human liposarcoma, prostate carcinoma, breast carcinoma, and gastric carcinoma(11, 12, 13, 14, 15). These results suggest that PPARγ activation may be implicated in the growth of malignant tumors.

There is a correlation between a high-fat diet and pancreatic carcinoma(1). The molecular link between dietary lipids and the tumor biology of pancreatic carcinoma is unknown. It has been shown that fat activates PPARγ. For instance, PPARγ modulates gene expression in response to fatty acid and lipid-derived metabolites(16, 17). This evidence led us to speculate that PPARγ may be implicated in the pathophysiology of human pancreatic carcinomas. In the present study, we therefore investigate the expression of PPARγ and examine the effects of PPARγ activation on cellular growth in human pancreatic adenocarcinoma cells.

All tissues used in this study were surgically resected specimens obtained at the Asahikawa Medical College Hospital. All pancreatic adenocarcinomas were of ductal cell origin (n = 10). Paired tumor and histologically normal pancreas were analyzed in each patient. For immunohistochemical analysis,formalin-fixed, paraffin-embedded blocks were obtained from tumors. Four-μm tissue sections were cut serially from each block.

Immunohistochemical staining for PPARγ was carried out by using the avidin-biotin complex method. Slides were deparaffinized, and endogenous peroxidase activity was blocked by incubation with 0.3%H₂O₂ in methanol for 10 min at room temperature. Sections were permeabilized in PBS-Triton and incubated in primary antibody for 3 h at room temperature. The rabbit polyclonal antibody to PPARγ (Calbiochem, San Diego, CA) was used at a 1:1000 dilution in PBS-Triton. Each section was then incubated with biotinylated goat antirabbit IgG for 2 h and with avidin-biotin-peroxidase complex for 1 h. Finally, the sections were reacted with 0.02% 3,3′-diaminobenzidine and 0.005%H₂O₂ in 0.05 mTris-HCl buffer. No immunostaining was observed when sections were incubated with the antibody in the presence of an excess amount of the antigenic PPARγ NH₂-terminal peptide used for immunization or with normal rabbit IgG. Only unequivocal nuclear staining was considered positive.

Human pancreatic carcinoma cell lines PK-1, PK-8, and PK-9 were obtained from Riken Cell Bank (Tokyo, Japan) and cultured in RPMI 1640(Life Technologies, Inc., Grand Island, NY) supplemented with 100 units/ml penicillin, 100 μg/ml streptomycin, 2.5 μg/ml amphotericin, and 10% fetal bovine serum. MIA Paca-2 was provided by the Japanese Cancer Research Resources Bank (Tokyo, Japan)and cultured in DMEM (Life Technologies, Inc.) supplemented with 100 units/ml penicillin, 100 μg/ml streptomycin, 2.5 μg/ml amphotericin, and 10% fetal bovine serum. Because PPARγ protein expression had already been reported in human colon cancer cell line HT-29 (6), we used this cell line as a positive control in experiments to detect PPARγ mRNA and protein expression. HT-29 was purchased from the American Type Culture Collection (Manassas, VA) and cultured in DMEM (Life Technologies, Inc.) containing 450 mg/dl glucose and supplemented with 2 mm l-glutamine, MEM nonessential amino acid solution (Sigma-Aldrich, Co., Irvine, United Kingdom), 100 units/ml penicillin, 100 μg/ml streptomycin, 2.5μg/ml amphotericin, and 10% fetal bovine serum. Cells were incubated at 37°C in a humidified atmosphere of 5% CO₂in air.

Troglitazone, a selective ligand for PPARγ, was kindly provided by Sankyo Pharmaceutical Co. (Tokyo, Japan) and dissolved in DMSO with a final concentration of 0.05% in the culture medium.

Total RNA was extracted from cultured cells using a modified version of the acid guanidinium thiocyanate/phenol/chloroform method using a single reagent [RNA-STAT 60; TelTest, Inc., Friendswood, TX (18, 19)]. Samples were dissolved with diethyl pyrocarbonate-treated water (RNase free). To remove contaminating genomic DNA, the RNA was treated with 10 μl of RQ1, RNase-free DNase (Promega, Madison, WI),0.5 μl of RNase inhibitor (Takara Shuzou Co., Shiga, Japan), and 10μl of 10× DNase buffer [400 mm Tris-HCl (pH 7.9), 100 mm NaCl, 60 mm MgCl₂, and 100 mm CaCl₂] in a final volume of 100 μl for 30 min at 37°C. RNA samples were purified by phenol-chloroform extraction and isopropanol precipitation. The resultant RNA samples were quantified using a spectrophotometer at a wavelength of 260 nm. The integrity of the isolated RNA samples was analyzed electrophoretically on an agarose gel, followed by staining in ethidium bromide.

An aliquot of 1 μg of total RNA from each sample was reverse transcribed to cDNA using the First-Strand cDNA Synthesis Kit(Pharmacia LKB Biotechnology, Uppsala, Sweden) according to manufacturer’s instructions with oligodeoxythymidylic acid primer. For detection of human PPARγ mRNA, a combination of a sense primer (5′-TCTCTCCGTAATGGAAGACC-3′) and an antisense primer(5′-GCATTATGAGACATCCCCAC-3′) was used as described previously(20). The amplification was carried out in a 100-μl mixture containing 1 μl of the above-mentioned cDNA product(corresponding to cDNA synthesized from 67 ng of total RNA), 0.4μ m each of the sense and antisense primers, 10 mm Tris-HCl (pH 8.3), 50 mm KCl, 1.5 mm MgCl₂, 200 μmdeoxynucleotide triphosphates, and 2.5 units of Taq DNA polymerase(Takara Shuzou Co.). The reaction conditions were as follows:(a) initial denaturation at 95°C for 2 min; (b)40 cycles of amplification (95°C for 40 s, 55°C for 50 s,and 72°C for 50 s); and (c) a final extension step of 7 min at 72°C. The PCR reaction products were separated electrophoretically in a 2% agarose gel and stained with ethidium bromide.

Ten μg of total RNA denatured in formamide and formaldehyde were electrophoresed through 1% formaldehyde-containing agarose gels as described previously (15). After electrophoresis, the RNA was transferred to a nylon membrane (Hybond N; Amersham International,Buckinghamshire, United Kingdom) by capillary blotting and then fixed by UV cross-linker (FUNA-UV-LINKER; Funakoshi, Tokyo, Japan). Prehybridization was performed at 42°C for 2 h in 50%formamide, 25 mm sodium phosphate (pH 6.5), 0.1% SDS, 5×SSC, 5× Denhardt’s solution, and 100 μg/ml denatured salmon sperm DNA. Hybridization was carried out at the same temperature for 20 h in the same solution with ³²P-labeled cDNA probes. The probe for PPARγ was amplified by PCR using MKN45 cell cDNA as a template and sequenced as described previously(15). β-Actin cDNA probe (Wako Chemicals Industries,Osaka, Japan) was used as an internal control. After washing the membrane under appropriately stringent conditions, the hybridization signals were analyzed with a bioimaging analyzer system (Fuji-BAS; Fuji Photo Film Co., Tokyo, Japan) or by autoradiography using XAR film(Eastman Kodak, Rochester, NY).

Transfections and luciferase assays were performed as described previously (15). PK-1 cells were seeded at a concentration of 1 × 10⁵ cells/35-mm dish and transfected with the plasmids 24 h after transfer to fresh media. Transfection was done using Lipofectin reagent (Life Technologies,Inc.) mixed with 2 μg of acyl-CoA oxidase promoter luciferase plasmid[kindly donated by Dr. T. Osumi, Himeji Institute of Technology,Hyogo, Japan (21)] and 0.2 μg of pRL-SV40(Promega) for 3 h. The transfection mixture was replaced by complete media with DMSO and 0.1, 10, or 100 μmtroglitazone and incubated for an additional 16 h. The cells were lysed with 1× luciferase lysis buffer (Toyo Ink, Tokyo, Japan). Luciferase activity was measured using the PicaGene reagent kit (Toyo Ink) in a luminometer (MiniLumat, Berthold, Widbad, Germany). The enzyme activity was normalized for efficiency of transfection on the basis of sea pansy luciferase activity, and relative values were determined. Transfection experiments were carried out independently two times, and the average values were calculated.

To evaluate the effect of PPARγ activation on cell growth, cells were seeded on a 96-well cell culture cluster (Corning Inc., Corning, NY) at a concentration of 1 × 10⁴cells/well in a volume of 100 μl. Twenty-four h later, each well was incubated with troglitazone at several concentrations. Cell numbers were measured colorimetrically using the Cell Counting Kit (Dojindo,Kumamoto, Japan) by ImmunoMini NJ-2300 (NJ InterMed, Tokyo, Japan)at a test wavelength of 450 nm. This assay is based on the cleavage of the 2-(4-iodophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium monosodium salt (WST-1) by mitochondrial dehydrogenase in viable cells(22). In comparison to the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay,WST-1 is more sensitive and does not require cells to be solubilized.

PK-1 cells treated with troglitazone or DMSO for 0 or 36 h were collected by centrifugation and permeabilized with ice-cold 70%ethanol for at least 1 h. After washing with PBS, cells were treated with PBS containing 100 mg/ml RNase A (DNase free) at 37°C for 30 min. After centrifugation, cells were resuspended in PBS containing 50 mg/ml propidium iodide and stained at 37°C for 30 min. DNA contents were analyzed by FACScan (Becton Dickinson).

The effect of troglitazone on the expression of CDK inhibitors p27^Kip1, p21^Cip1/Waf1, and p18^Ink4c was studied in PK-1 cells by Western blot analysis. PK-1 cells were treated with several doses of troglitazone, and total proteins were extracted from PK-1 cells at several time points. Protein concentrations were measured using Bio-Rad Protein Assay Reagent (Bio-Rad, Richmond, CA) following the manufacturer’s suggested procedure. Fifty μg of protein were separated by 5–20% SDS-PAGE (Ready Gels J; Bio-Rad). After electrophoresis, the proteins were transferred to nitrocellulose membrane (Amersham Life Science, Inc., Piscataway, NJ), blocked overnight in TBS with 10% skim milk at 4°C, reacted with primarily polyclonal antibody against human p27^Kip1 or p18^Ink4c (Santa Cruz Biotechnology, Santa Cruz,CA) or mouse anti-p21^Cip1/Waf1 monoclonal antibody (PharMingen International, Tokyo, Japan), and washed. After reaction with horseradish peroxidase-conjugated antigoat IgG or antimouse IgG, immune complexes were visualized by using the ECL detection reagents (Amersham International) following the manufacturer’s suggested procedure. Normal goat or mouse IgG was used simultaneously as a control.

The sequence for the antisense oligonucleotide targeted against p27^kip1 was 5′-UGGCUCUCCUGCGCC-3′, and the mismatch oligonucleotide was 5′-UCCCUUUGGCGCGCC-3′, as described previously (23). Both oligonucleotides were labeled with fluorescent FITC. Before the addition of oligonucleotides to exponentially growing cells on tissue culture plates, PK-1 cells were replenished with OPTI-MEM (Life Technologies, Inc.). Cells were then treated with 5 μm oligonucleotides in the presence of 50 nm Lipofectin (Life Technologies, Inc.). After 3 h of incubation at 37°C, oligonucleotides were taken up by more than 80%of the cells, as determined by fluorescence microscopy (data not shown). The medium was then removed and replaced with fresh complete medium. The cells were incubated for an additional 10 h and then harvested with trypsin and plated into a 96-well cell culture cluster for WST-1 assay or into 60-mm dishes for Western blot analysis. The cells were incubated with either troglitazone at a concentration of 100μ m or DMSO. Cell numbers and p27^Kip1 protein levels were evaluated as described above.

The results were expressed as the mean ± SE. Statistical analysis was performed by one-way ANOVA and subsequent Fisher’s LSD test. Ps of <0.05 were considered statistically significant.

Fig. 1 shows the representative immunohistochemical findings of the surgically resected pancreas tissue. Fig. 1,A shows a representative sample containing normal duct epithelial cells. In these normal duct cells, PPARγ protein was not detected as shown in Fig. 1,B. Staining for PPARγ was negligible in nontumorous duct epithelial cells of the pancreas in all cases. In contrast, Fig. 1, C—E demonstrates that PPARγ protein was clearly present in the nuclei of the pancreatic adenocarcinoma cells in three representative cases. PPARγ protein was detected in the nuclei of the pancreatic cancer cells in 9 of 10 adenocarcinomas of the pancreas. In all nine of the cases positive for PPARγ, more than 80%of carcinoma cells were stained.

We examined whether PPARγ mRNA was expressed in cultured human pancreatic carcinoma cells. Fig. 2 demonstrates the results of PPARγ gene expression. RT-PCR(Fig. 2,A) and Northern blot analysis (Fig. 2 B)clearly showed that PPARγ mRNA is expressed in all of the tested pancreatic carcinoma cell lines [MIA Paca-2 (Lane 2), PK-1(Lane 3), PK-8 (Lane 4), and PK-9 (Lane 5)]. Among these four pancreatic carcinoma cell lines, Northern blot analysis showed that PK-1 expresses the highest level of PPARγmRNA. We also detected PPARγ mRNA expression in HT-29 (Lane 1), a colon carcinoma cell line that is known to express a large amount of PPARγ mRNA. When compared with the mRNA expression in HT-29, the PPARγ mRNA expression, even in PK-1 cells, was relatively weaker.

PK-1 cells were transfected with an acyl-CoA oxidase promoter luciferase reporter plasmid containing a PPRE. Treatment with troglitazone, a selective ligand for PPARγ (24), for 12 h increased the luciferase activity in the PK-1 cells in a dose-dependent fashion (Fig. 3). The luciferase activity seen with a 100 μm dose of troglitazone was approximately 15 times higher than that seen with vehicle only.

Next we examined the effect of PPARγ activation on cell growth in pancreatic carcinoma cells that express PPARγ as shown in the above-mentioned study. Fig. 4 illustrates the effect of troglitazone at several doses on cell growth in PK-1, PK-8, PK-9, and MIA Paca-2 cells. Troglitazone significantly inhibited the growth of PK-1 cells in a dose-dependent manner. A 100μ m dose of troglitazone completely blocked cell growth in PK-1 cells during the tested time period. Similarly, cell growth of PK-8, PK-9, and MIA Paca-2 was dose-dependently inhibited by troglitazone.

To investigate the mechanism of growth inhibition by PPARγactivation, we examined the effect of troglitazone on cell cycle progression. PK-1 cells were treated with a 100 μm dose of troglitazone for 48 h. PK-1 cells were then stained with propidium iodide, and DNA content was determined by flow cytometry. Cell cycle profiles of PK-1 cells are shown in Fig. 5. No change in the percentage of cells in G₁ or S phase was observed in PK-1 cells treated with vehicle for 36 h(G₁, 55.8%; S phase, 37.9%) when compared with the percentage of cells in G₁ or S phase in cells treated for 0 h (G₁, 54.1%; S phase,37.3%). The population of G₁ or S-phase cells in PK-1 cells treated with troglitazone for 36 h was 61.4% and 31.4%, respectively, indicating an increase in the number of cells in G₁ phase and a reduction in the number of cells in S phase by troglitazone.

Fig. 6 shows a representative Western blot analysis for p27^Kip1 in PK-1 cells treated with troglitazone. The time course study (Fig. 6,A) clearly illustrates that troglitazone at a dose of 100 μm did not change the p27^Kip1 protein level at 6 h (Lane 2) but increased the p27^Kip1 protein level at 12, 24, and 36 h (Lanes 3–5) when compared with that at 0 h (Lane 1). Fig. 6,B illustrates the dose-response effect of 36 h of troglitazone treatment on p27^Kip1 protein levels in PK-1 cells. As shown,troglitazone dose-dependently increased p27^Kip1protein levels. In contrast, troglitazone at a dose of 100μ m failed to increase the protein level of p21^Cip1/Waf1 and p18^Ink4c(Fig. 6 C).

To determine whether an increase in p27^Kip1 was responsible for the troglitazone-induced growth inhibition in pancreatic carcinoma cells, we used an antisense oligonucleotide to down-regulate p27^Kip1. As shown in Fig. 7, the p27^Kip1 protein level was increased 24 h after troglitazone treatment in mismatch oligonucleotide-treated PK-1 cells, in agreement with the data on nontransfected PK-1 cells as described above. The p27^Kip1 protein level in antisense-treated PK-1 cells 24 h after troglitazone treatment was increased when compared with that before troglitazone treatment but was smaller than that in mismatch oligonucleotide-treated cells at the same time period. These findings indicate that the antisense oligonucleotide down-regulated p27^Kip1 protein expression.

We next examined the effect of troglitazone on cell proliferation in the antisense-treated cells. Troglitazone at a dose of 100μ m inhibited cell proliferation in the mismatch oligonucleotide-treated cells (Fig. 8,A). In contrast, the same dose of troglitazone failed to induce growth inhibition in the antisense-treated cells (Fig. 8 B).

Increasing evidence suggests that PPARγ may be implicated in cell differentiation (3, 4). It has been reported thus far that PPARγ expression and the effects of PPARγ ligands are involved in cell proliferation in colon carcinoma, breast carcinoma, prostate carcinoma, gastric carcinoma, and liposarcoma (6, 7, 8, 9, 10, 11, 12, 13, 14, 15). However, little is known regarding whether human pancreatic carcinoma expresses PPARγ. In addition, there is no evidence of the effects of PPARγ activation on cell behavior in pancreatic carcinoma cells. In the present study, we examined the above-mentioned questions using surgically resected human pancreatic adenocarcinomas and human pancreatic carcinoma cell lines.

First of all, the present study clearly demonstrates for the first time that human pancreatic adenocarcinoma cells express PPARγ. RT-PCR and Northern blot analysis demonstrated PPARγ mRNA in all four human pancreatic carcinoma cell lines tested. Immunohistochemical study in surgically resected tissues also showed PPARγ protein in the nuclei of carcinoma cells in 9 of 10 pancreatic adenocarcinomas. These results strongly indicate that a majority of human pancreatic adenocarcinoma cells express PPARγ. In contrast, PPARγ protein was not detected immunohistochemically in the nuclei of normal pancreatic duct epithelial cells. Thus, PPARγ expression may be up-regulated in pancreatic carcinoma cells, although we have not compared PPARγ mRNA levels in pancreatic carcinoma with those in normal pancreatic duct cells.

Next we performed transient transfection assays in PK-1 cells using a PPAR-responsive element cloned upstream of luciferase to determine whether the PPARγ expressed in pancreatic cancer cells is functional as a transcriptional factor. Luciferase assay in PK-1 cells showed that troglitazone, a selective ligand for PPARγ, transactivated the transcription of a PPRE-driven promoter in a dose-dependent fashion,indicating that PPARγ expressed in PK-1 cells is indeed functional as a transcriptional activator.

The effect of PPARγ activation on cellular growth in human pancreatic carcinoma cells was then examined, and this study has clearly revealed that troglitazone significantly inhibited the cell growth of all four pancreatic carcinoma cell lines tested in a dose-dependent manner. These results indicate that PPARγ activation has an antiproliferative action on pancreatic carcinoma cells. According to the dose-response effects of troglitazone on cell growth in this study, troglitazone at a dose as small as 1 μm could induce growth inhibition in PK-1 cells. The effective dose range of troglitazone to inhibit cell growth in PK-1 cells in this study is the same or less as that needed to induce growth inhibition in human colon carcinoma cells as shown by Saffaf et al. (6). Saffaf et al. have demonstarted that troglitazone inhibited colon carcinoma cell growth in vitro and that the growth of transplanted colon carcinoma cells was suppressed when mice were treated with troglitazone (6). The effects of troglitazone occurred at concentrations of ligand that are achieved in the treatment of human type 2 diabetes with troglitazone (25). This evidence suggests that troglitazone at doses used clinically may induce growth inhibition of colon carcinoma cells. The same dose range of troglitazone inhibited cell growth in colon carcinoma and pancreatic carcinoma cells in vitro, suggesting that antitumor action of troglitazone for pancreatic carcinoma might be given at a clinically relevant doses of troglitazone. This evidence encourages us to believe that PPARγ could be considered as a possible target molecule for the treatment of human pancreatic carcinomas.

The growth inhibition of pancreatic carcinoma cells by PPARγactivation observed in this study may explain a mechanism by which NSAIDs induce growth inhibition in human pancreatic carcinoma cells. Molina et al. (26) have recently demonstrated an increased expression of cyclooxygenase-2 in human pancreatic carcinomas and an inhibition of pancreatic carcinoma cell growth by NSAIDs. We agree that NSAIDs act as a cyclooxygenase inhibitor to inhibit the growth of pancreatic carcinoma cells. However, other mechanisms of inhibiting pancreatic carcinoma cell growth cannot be ruled out. Lehmann et al. (27) have shown that NSAIDs are known to be weak agonists of PPARγ. Accordingly, one cannot exclude the possibility that PPARγ activation may contribute,at least in part, to the growth inhibition in pancreatic carcinoma cells by NSAIDs.

To clarify the mechanism by which troglitazone inhibits the growth of pancreatic carcinoma cells, we analyzed the cell cycle profile in cells treated with troglitazone. According to our data obtained by flow cytometry, troglitazone increased the population of cells in G₁ phase and reduced the population of cells in S phase in PK-1 cells. These results suggest that the growth inhibition by troglitazone was induced by G₁ arrest. With regard to this point, earlier researchers have demonstrated that thiazolidinediones including troglitazone inhibit the growth of colon carcinoma cells and myeloid leukemia cells through the induction of G₁ arrest (7, 8, 28). The present result obtained in pancreatic carcinoma cells therefore supports the recent observations in other growing cells that PPARγ activation induces growth inhibition by G₁ cell cycle arrest. However, little is known about the detailed mechanism of troglitazone-induced G₁ arrest in any type of cells.

Multiple cyclins and CDKs are positive regulators of cell cycle progression. Cyclin/CDK complexes are negatively regulated by various CDK inhibitors (29, 30, 31, 32, 33). CDK inhibitors belong to two large families based on their structural and functional properties. The INK4 family, which includes p16^Ink4a,p15^Ink4b, p18^Ink4c, and p19^Ink4d, consists of tandem repeats of an ankyrin-like sequence, whereas the Cip/Kip family, which includes p21^Cip1/Waf1, p27^Kip1, and p57^Kip2, has a homologous NH₂-terminal domain that contains contiguous cyclin and CDK binding regions. p27^Kip1 is associated with cyclin D/CDK4, but it has the ability to inhibit various cyclin/CDK complexes in vitro (33, 34). p27^Kip1 mediates G₁ arrest induced by transforming growth factor β, contact inhibition, or serum deprivation in epithelial cell lines (32). In the present study, we examined the hypothesis that p27^Kip1may be involved in the G₁ arrest induced by troglitazone in PK-1 cells. Time course and dose-response studies clearly demonstrated that troglitazone potently increased p27^Kip1 protein levels in dose- and time-dependent fashions. In the present study, we also examined the protein levels of other CDK inhibitors, p18^Ink4cand p21^Cip1/Waf1, by Western blot analysis and showed that neither p18^Ink4c nor p21^Cip1/Waf1 protein level was increased by troglitazone treatment. These results suggest that p27^Kip1 might be a specific CDK inhibitor mediating the troglitazone-induced G₁ arrest in PK-1 cells.

Finally, we tried to clarify the hypothesis that an increased p27^Kip1 level was responsible for the troglitazone-induced growth inhibition in human pancreatic carcinoma cells. To down-regulate the p27^Kip1 protein level, we used an antisense oligonucleotide to p27^Kip1 according to previous reports (23, 35). Pretreatment with antisense p27^Kip1oligonucleotide reduced the increase in the p27^Kip1 protein level seen after troglitazone treatment and blocked troglitazone-induced growth inhibition. These result strongly suggest that p27^Kip1 plays a key role in the growth inhibition by troglitazone in human pancreatic carcinoma cells. Although we do not know the molecular mechanism by which p27^Kip1 protein levels are increased by troglitazone in PK-1 cells, it is generally accepted that the abundance of p27^Kip1 protein is regulated via translational(36, 37) and posttranslational pathways (38)and, less commonly, at the level of transcription (34). Rao et al. (39) have shown that lovastatin, a hydroxymethyl-glutaryl-CoA reductase, arrests cells by inhibiting the proteasome, which results in the accumulation of p27^Kip1, leading to G₁arrest. Thus, the mechanism by which lovastatin inhibits cell growth seems to be similar to the growth inhibition coupled with G₁ arrest and increased p27^Kip1 protein levels seen in PK-1 cells treated with troglitazone as demonstrated in the present study. It is therefore possible that troglitazone exerts an antiproliferative action by inhibiting the proteasome. Additional studies are needed to clarify this speculation.

In summary, the present study demonstrated for the first time that human pancreatic carcinoma cells express PPARγ, that PPARγexpression is detected in the nuclei of pancreatic carcinoma cells, and that PPARγ activation by troglitazone inhibited cellular growth. Moreover, an increased amount of p27^Kip1, a CDK inhibitor, may be involved in the G₁ arrest followed by growth inhibition by troglitazone in pancreatic carcinoma cells. Based on this evidence, we suggest that PPARγ should be listed as a target molecule for the treatment of human pancreatic carcinomas.