Peroxisome proliferator-activated receptor γ ligand-induced growth inhibition of human hepatocellular carcinoma

Peroxisome proliferator-activated receptor γ (PPARγ) ligands have been implicated in the growth inhibition and differentiation of certain human cancers with diverse tissue origin. In this study, expression of PPARγ in human hepatocellular carcinoma (HCC) and the effect of PPARγ ligands on HCC cells were investigated in vitro using Hep G2, HuH-7, KYN-1 and KYN-2 cell lines. All cell lines were found to express functionally active PPARγ and a marked growth inhibition was induced by thiazolidinedione ligands troglitazone, and pioglitazone as well as with its natural ligand 15-deoxy-Δ12,14-prostaglandin J 2. The growth inhibitory effect was associated with a dose-dependent inhibition of DNA synthesis, cell cycle progression and α fetoprotein expression. © 2001 Cancer Research Campaign http://www.bjcancer.com

Peroxisome proliferator-activated receptor γ (PPARγ), a member of the nuclear hormone receptors, can act as ligand-sensitive transcription factor (Mangelsdorf et al, 1995;Kliewer and Willson, 1998). Activated receptors heterodimerize with retinoid X receptor (RXR) and can alter transcription of target genes after binding to peroxisome proliferator responsive elements (PPRE) (Kliewer et al, 1992). PPARγ was initially reported for its regulatory roles in insulin sensitization and adipocyte differentiation (Chawla et al, 1994;Tontonoz et al, 1994aTontonoz et al, , 1994b. However, later studies have shown that PPARγ is also expressed in other cell types and it has recently been of interest for its role in cell proliferation and cancer. In vitro studies have revealed the growth inhibitory effects of PPARγ ligands on human cancer cells of different tissue origin, including liposarcoma (Tontonoz et al, 1997), adenocarcinoma of breast (Elstner et al, 1998;Mueller et al, 1998;Clay et al, 1999), colorectal adenocarcinoma and carcinoma Brockman et al, 1998;Kitamura et al, 1999), gastric adenocarcinoma (Takahashi et al, 1999;Sato et al, 2000), pancreatic carcinoma , adenocarcinoma and carcinoma of prostate (Kubota et al, 1998;Butler et al, 2000), transitional epithelial cancer of urinary bladder (Guan et al, 1999), chorio carcinoma of placenta (Keelan et al, 1999), carcinoma of lung and non-small cell lung cancer (Chang and Szabo, 2000;Tsubouchi et al, 2000) and myeloid leukaemias (Asou et al, 1999;Hirase et al, 1999;Sugimura et al, 1999). In vivo study on xenograft of human tumours in immunodeficient mice followed by troglitazone treatment also provided similar results (Elstner et al, 1998;Kubota et al, 1998;Sarraf et al, 1998). Ligand-mediated PPARγ activation in those cancer cells induced cell cycle arrest (Tontonoz et al, 1997;Brockman et al, 1998;Sugimura et al, 1999;, differentiation (Tontonoz et al, 1997;Kubota et al, 1998;Mueller et al, 1998;Sarraf et al, 1998;Chang and Szabo, 2000), and apoptosis (Elstner et al, 1998;Clay et al, 1999;Keelan et al, 1999;Sato et al, 2000) or nonapoptotic cell death (Kubota et al, 1998;Butler et al, 2000). Recently, troglitazone has been used in clinical trial for the patients with advanced liposarcoma (Demetri et al, 1999) and for patients with advanced prostate cancer (Hisatake et al, 2000;Mueller et al, 2000). The drug was found to induce histologic and biochemical differentiation in liposarcoma and a prolonged stabilization of prostate-specific antigen (PSA) level in prostate cancer patients. Such results suggest that ligands of PPARγ may serve as a biological modifier in human cancers and the therapeutic potential should be further investigated. Human liver tissue has been reported to express PPARγ (Auboeuf et al, 1997;Vidal-Puig et al, 1997), however, expression of PPARγ in human hepatocellular carcinoma (HCC) and the effect of PPARγ agonists have not yet been studied and this study was designed to investigate that.
PPARγ can be activated by certain polyunsaturated fatty acids Xu et al, 1999), prostaglandin J 2 metabolite 15-deoxy-∆ 12,14 -PGJ 2 (15d-PGJ 2 ) (Forman et al, 1995;, the thiazolidinedione (TZD) class of antidiabetic drugs Elbrecht et al, 1996) and a variety of nonsteroidal anti-inflammatory drugs . Recent attention has focused on troglitazone because of its rare but potentially lethal hepatotoxicity (Watkins and Whitcomb, 1998;Kohlroser et al, 2000) although at present, no similar side effects have been observed with other TZD, either rosiglitazone or pioglitazone (Henney, 1999). Because evidence to date does not indicate that hepatotoxicity is attributable to TZDs as a class or to PPARγ agonists in general, in vitro growth inhibitory effects of PPARγ Hepatocellular carcinoma cell lines Hep G2 was obtained from American Type Culture Collection (Manassas, VA, USA), HuH-7 from JCRB Cell Bank (National Institute of Health Sciences, Osaka, Japan) whereas, KYN-1 and KYN-2 were kind gifts from Prof M Kojiro (Department of Pathology, Kurume University School of Medicine, Japan). Hep G2 (Aden et al, 1979), as well as HuH-7 (Nakabayashi et al, 1982) was established from well differentiated hepatocellular carcinoma, KYN-1 (Yano et al, 1986) from a moderately differentiated, and KYN-2 from a pleomorphic hepatocellular carcinoma corresponding to Edmondson-Steiner grade III (Yano et al, 1988). All the hepatocellular carcinoma cell lines secrete albumin and AFP, in addition, KYN-1 possesses the nature of transformation to adenocarcinoma with production of mucicarmin-positive materials (Yano et al, 1986). Cells were grown at 37˚C in Dulbecco's Modified Eagle Medium (DMEM) (Nissui Pharmaceutical, Tokyo, Japan) supplemented with 10% fetal bovine serum (ICN Biomedicals, Aurora, Ohio, USA), L-glutamine (ICN Biomedicals), penicillin-streptomycin (ICN Biomedicals) and maintained in an incubator with 5% CO 2 and constant humidity.
Sensitivity to PPARγ ligands was studied in all the cell lines. However, changes in cell cycle distribution caused by PPARγ ligands, as well as protein level of cyclin-dependent kinase (CDK) inhibitors and α fetoprotein (AFP) or albumin mRNA expression after troglitazone treatment were investigated using the representative cell line Hep G2.
A dose ranging from 0.1 to 50 µM troglitazone was selected for this in vitro study according to the pharmacokinetics (Plosker and Faulds, 1999) and tissue distribution of the drug (Kawai et al, 1997). For comparison of the dose effect, the 2 other PPARγ ligands, pioglitazone and 15d-PGJ2 were also applied to cells at the same dosages.

RT-PCR
PPARγ expression at mRNA level was investigated by RT-PCR. From each cell line, total RNA was extracted by Isogen (Nippon Gene, Tokyo, Japan). 7.5 µg of total RNA was reverse transcribed with oligo dT primer in a 50 µl reaction using ProstarTM First-Strand RT-PCR kit (Stratagene Cloning Systems, La Jolla, CA, USA). 1 µl of the cDNA was amplified by PCR on GeneAmp PCR system 9600 (PE Applied Biosystems, Foster City, CA, USA). Amplification was carried out in a 25 µl reaction volume using 10 pmol of each of the primers (sense-5′TCTGGCCCAC-CAACTTTGGG 3′ and antisense-5′ CTTCACAAGCATGAACT-CCA 3′) (Kubota et al, 1998) with 200 µM each dNTP, 1.5 mM MgCl 2 and 0.75 unit of Amplitaq gold (PE Applied Biosystems) for 30 cycles (94˚C 30 s, 56˚C 30 s and 72˚C 60 s). 5 µl of PCR products were electrophoresced on 1.5% agarose gel with 0.5% ethidium bromide and visualized on UV. As internal control, βactin cDNA sequence was amplified using the human β-actin control amplimer set 5′ ATCTGGCACCACACCTTCTACAAT-GAGCTGCG 3′ (sense) and 5′ CGTCATACTCCTGCTTGCT-GATCCACATCTGC 3′ (antisense) (Clontech Laboratories, Palo Alto, CA, USA). In all cases RNA processed without reverse transcriptase and preparation without template was used to check carryover and amplified products purified by Geneclean (Bio 101, Carlstad, CA, USA) were confirmed by direct sequencing on ABI prism 310 Genetic Analyser (PE Applied Biosystems) using BigDye Terminator Cycle Sequencing kit (PE Applied Biosystems).

PPARγ functional assay
To investigate whether PPARγ expressed in HCC cell lines are functionally active, we measured the reporter gene activity using luciferase assay (Alam and Cook, 1990;Brasier et al, 1989;Nordeen, 1988). Induction of the transcriptional activation by PPARγ ligand troglitazone through PPRE (Kliewer et al, 1992;Juge-Aubry et al, 1997) was assessed by the luciferase assay after transfecting Hep G2 cells a eukaryotic expression vector of firefly luciferase with PPRE cloned upstream to its SV40 promoter. PPRE sequence from the acyl-CoA oxidase promoter was synthesized as described . Oligonucleotides 5′ GATCCGGACCAGGACAAAGGTCACGTTCGGACCAGGA-CAAAGGTCACGTTCGTCCTQQTCCQ3′ and 5′GATCCGAA-CGTGACCTTTGTCCTGGTCCGAACGTGACCTTTGTCCTG GTCCG 3′ were annealed and 2 copies were cloned into the multiple cloning site (MCS) upstream to SV40 promoter of the PicaGene promoter vector (PGV-P2) (Nippon Gene) to construct PGVP2.PPRE. Standard dual luciferase assay was performed in 24-well plates, transfecting 10 5 cells well -1 with 1 µg of PGVP2.PPRE and 0.05 µg of PicaGene SeaPansy TK control vector (pRL-TK) (Nippon Gene) using DOTAP liposomal transfection reagent (Roche Molecular Biochemicals, Mannheim, Germany). 10 h after the transfection, fresh media containing 10 µM troglitazone or vehicle only was added to the transfected cells.
After 12 h treatment with troglitazone, cells were lysed with passive lysis buffer and luciferase assay was performed with PicaGene Dual SeaPansy Luminescence Kit (Nippon Gene) on Lumat LB9506 luminometer (EG & G Berthhold, Bad Wildbad, Germany). Results were evaluated after normalization of the firefly luciferase activity of PGVP2.PPRE with that of sea pansy luciferase activity of pRL-TK. Tests were performed in duplicate, repeated twice and the mean value of increment in luciferase activity were analysed.

Assessment of growth inhibition
Viable cell counting Effect on cell growth was evaluated by direct cell counting using a haemocytometer. 5 × 10 5 cells were seeded into 60 mm plates and after 24 h, PPARγ ligand troglitazone, pioglitazone or 15d-PGJ 2 was added to culture media in 10, 25 and 50 µM concentration. After 48 h, both floating and adherent cells were harvested and viable cells were counted by trypan blue dye exclusion. Relative rate of increment in cell counts in presence of ligands compared with that of control without drug was considered representative of cell growth.
Assay for DNA synthesis DNA synthesis was assessed by 3 H-thymidine incorporation. 10 5 cells were seeded into 24-well plates and after 24 h culture in complete media, experimental drugs in varying concentrations dissolved in media was added to the growing cells. After 24 h treatment with drug, 1 µCi [methyl-3 H]-thymidine (Amersham Pharmacia Biotech) was added to each well and incubated for a further 6 h. Then the cells were trypsinated and harvested onto a glassfibre filtermat by a cell harvester, dried 1 h and 3 H-thymidine incorporation was measured with 1450 Microbeta TM scintillation counter (Wallac Oy, Turku, Finland). Assays were performed in triplicate, the mean CPM values after normalization were analysed for relative 3 H-thymidine incorporation and repeated thrice.

Cell cycle analysis
Cell-cycle analysis was performed on the representative Hep G2 cells. Approximately 2 × 10 6 cells were seeded in 100 mm plates and on growing cells, fresh media with experimental drugs in various concentrations was added. Depending on the results of viable cell counts and 3 H-thymidine incorporation assay results, troglitazone and pioglitazone was used in 10, 25 and 50 µM and 15d-PGJ 2 in 5, 10 and 25 µM. After 48 h drug treatment, cells were harvested, washed with PBS and fixed overnight in cold (-20˚C) 75% ethanol. Ethanol fixed cells were washed twice with ice cold PBS, and then approximately equal number of control or drug treated cells were treated with 200 µg ml -1 RNase for 60 min at 37˚C and stained with 100 µg ml -1 propidium iodide 30 min at room temperature in dark. Cellular DNA contents were analysed by flow cytometry (EPICS Elite, Coulter Electronics, FL, USA). In each case, histogram of DNA contents in 10 000 cells were analysed using Multicycle AV software (Phoenix, San Diego, CA, USA) to evaluate relative distribution of cells in G1, S and G2 phase. Cell cycle analysis on Hep G2 cells was performed in duplicate and repeated 3 times for each drug.

Northern blot analysis for albumin and AFP expression
Whether troglitazone induced differentiation in hepatocellular carcinoma cells, was investigated after 2 × 48 h pulse exposure of Hep G2 cells with troglitazone in varying concentrations. Extracted total RNA was used in Northern blot analysis for the quantitative comparison in the expression of albumin and AFP. Briefly, 10 µg of total RNA was electrophoresced in formaldehyde-containing 1% agarose gel and transferred to nitrocellulose membrane (Hybond-C Extra, Amersham Pharmacia Biotech), followed by UV cross-linking (Funa-UV-linker, Funakoshi, Tokyo, Japan). 32 P-labelled (Multiprime DNA labelling systems, Amersham Pharmacia Biotech) probes used for the Northern blot analysis were a 422 bp cDNA fragment of human albumin (Niwa et al, 1996), a 395 bp cDNA fragment of human AFP (Niwa et al, 1996) and the 838 bp cDNA fragment of human β-actin (RT-PCR, internal control). After 4 h prehybridization, hybridization was carried out for 20 h at 42˚C and then membranes were washed twice at 37˚C in 2 × SSC containing 0.2% SDS for 10 min and once at 50˚C in 0.1 × SSC containing 0.2% SDS for 30 min. Autoradiography was performed using FUJIX BAS-2000 II (Fuji Photo Film, Tokyo, Japan) and photographed with a FUJIX Pictrography 3000 (Fuji Photo Film). The relative intensities of message bands were quantified using the FUJIX-BAS 2000 II Image Analyser.

Statistical analysis
Data were expressed as mean ± SE. Values were compared and significant differences between means were determined by Analysis of Variance (ANOVA). Multiple comparisons were done by Least Significant Difference (LSD) test after ANOVA. P values < 0.05 were considered significant.

HCC cell lines expressed PPARγ
RT-PCR analysis readily detected the expression of PPARγ mRNA in HCC cell line Hep G2, HuH-7, KYN-1 and KYN-2 ( Figure 1A). Amplified fragment was appropriate for 360 bp (Kubota et al, 1998) and direct sequencing also confirmed that RT-PCR products were identical to human PPARγ cDNA sequence (Elbrecht et al, 1996). Western blot analysis with anti-human PPARγ monoclonal antibody also detected expression of PPARγ in all HCC cell lines ( Figure 1B) with molecular mass of approximately 54 kDa (Elbrecht et al, 1996).

Expressed PPARγ was functional
Transactivation experiment showed that 10 µM troglitazone induced about 2.2 to 3.8 fold enhanced luciferase activity ( Figure  1C) in all the HCC cell lines transfected with a luciferase reporter vector containing PPRE upstream to SV40 promoter. Although induction of luciferase activity was relatively lower in KYN-1 and KYN-2, the differences were not significant.

PPARγ ligands induced growth inhibition
Trypan blue staining and direct counting of viable cells 48 h after troglitazone treatment at 10, 25 and 50 µM concentration showed a significant decrease in relative increment cell number in all cell lines (Figure 2A). With equivalent dose, 15d-PGJ 2 also reduced the viable cell count of Hep G2 significantly ( Figure 2B). However, at higher concentration (i.e. 25 and 50 µM) of 15d-PGJ 2 and troglitazone, cell death was marked. Pioglitazone induced inhibition of cell growth was quantitatively less than that of troglitazone and 15d-PGJ 2 as observed on Hep G2 cells ( Figure 2B).

PPARγ ligands inhibited DNA synthesis
Troglitazone induced inhibition of DNA synthesis as evidenced by 3 H-thymidine incorporation in all the HCC cell lines after 30 h drug treatment. There was a significant dose-dependent decrease in 3 H-thymidine incorporation with all concentrations of troglitazone used starting from 1 µM in Hep G2, HuH-7 and KYN-1 but 5 µM in KYN-2 ( Figure 3A). More than 95% inhibition was observed in all cell lines at 50 µM troglitazone, about 80% at 25 µM and nearly 50% at 10 µM ( Figure 3A). Pioglitazone induced inhibition of 3 H-thymidine incorporation was found significant from 1 µM onward in Hep G2, KYN-1 and KYN-2 but 5 µM in HuH-7 ( Figure 3B). However, 15d-PGJ 2 mediated significant inhibition at 1 µM in KYN-2 cells and from 5 µM in other cell lines ( Figure 3C).

PPARγ ligands altered cell cycle progression
The cell cycle distribution changes were evident after exposure of Hep G2 cells to troglitazone, pioglitazone and 15d-PGJ 2 ( Figure  4). With increasing dose of the ligands, cell cycle distribution significantly increased in G1 phase and decreased in S phase (Figure 4). In addition, 15d-PGJ 2 also induced a significant increase in G2 phase ( Figure 4C). Although G2 phase increased slightly with troglitazone ( Figure 4A), the dose-dependent G2 changes with troglitazone or pioglitazone were found to be statistically insignificant.

Troglitazone increased p27 Kipl and p18 Ink4c protein levels
Quantitative comparison of dose-dependent changes in the CDK inhibitor levels in Hep G2 cells after troglitazone treatment is shown in Figure 5 (A, B and C). Cell cycle arrest evident at 48 h troglitazone treatment was associated with a dose-dependent increase in CDK inhibitor p27 Kipl and p18 Ink4c , with a decline in p21 Cipl/Wafl level.

Troglitazone decreased AFP expression
Northern blot analysis on RNA extracted from Hep G2 cells after exposure to troglitazone, revealed that expression of AFP decreased while there was a little increase in the expression of albumin in 5 and 10 µM but decreased at 25 µM ( Figure 6A). On repeated experiments, it was found that even with a decreased albumin expression at 25 µM troglitazone, the ratio of albumin versus AFP expression increased consistently in a dose-dependent way ( Figure 6B).

DISCUSSION
Drugs acting through PPARγ have growth inhibitory effects on certain human cancers. Both thiazolidinedione and nonthiazolidinedione ligands have been shown to exert the inhibitory effect.
In this study, we evaluated the initial aspects of growth inhibitory potential of PPARγ agonists on human HCC cell lines using the commonly studied TZD ligand troglitazone, a nonhepatotoxic TZD pioglitazone and the endogenous ligand 15d-PGJ 2 . We found that human HCC cell lines Hep G2, HuH-7, KYN-1 and KYN-2 express PPARγ and the expressed PPARγ was functionally active. PPARγ ligands also induced growth inhibition in all the HCC cell lines. Compared to the rate of cell-count increment in drug-free control, troglitazone as well as pioglitazone and 15d-PGJ 2 caused a significant inhibition in cell growth. Inhibition of cell proliferation was also evident in 3 H-thymidine incorporation assays indicating that PPARγ ligands induced inhibition of DNA synthesis in HCC cell lines. A similar dose response in inhibition of thymidine incorporation was also reported on colon cancer cells (Kitamura et al, 1999).
Regarding the dose response of troglitazone in this study, a significant growth inhibition could be obtained at 1 µM in 3 of the 4 cell lines, with marked inhibition in all at 5 µM. The other 2 PPARγ ligands also exhibited similar dose responses. According to the pharmacokinetics of troglitazone (Ploskar and Faulds, 1999) and due to a higher tissue distribution in liver (Kawai et al, 1997), a daily dose of 600 to 800 mg troglitazone may attain an effective drug level in vivo. Similarly, 30 to 60 mg daily dose of pioglitazone would result in an effective drug level. In a recent clinical trial on prostate cancer patients, 800 mg daily dose of troglitazone has already been used (Mueller et al, 2000).
Although the exact mechanism of growth inhibition of tumour cells by PPARγ ligands is not well-understood (Gelman et al, 1999), it was reported to be associated with alteration in the cell cycle. Activation of PPARγ resulted in G1 cell cycle arrest in colon cancer cells (Brockman et al, 1998;Kitamura et al, 1999), pancreatic cancer cells  and leukaemia cells (Asou et al, 1999;Sugimura et al, 1999). We found that all PPARγ ligands dose dependently increased Hep G2 cells accumulating in G1 phase and decreased in S phase. Moreover, 15d-PGJ 2 and to a lesser extent troglitazone, induced also G2 cell cycle arrest. A G2 arrest in addition to G1 cell cycle arrest may be responsible for the higher growth inhibitory potential of 15d-PGJ 2 and troglitazone over pioglitazone.
To reveal the underlying mechanism of troglitazone-induced cell cycle arrest, CDK inhibitor p27 Kipl , p21 Cipl/Wafl and p18 Ink4c protein levels were studied in Hep G2 cells. A dose-dependent increase in p27 Kipl and p18 Ink4c with decreased p21 Cipl/Wafl level was observed 48 h after troglitazone treatment. The exact mechanism of cell cycle withdrawal induced by PPARγ ligands is not yet clear but it was found related to up-regulation of p21 Cipl/Wafl in eosinophilic leukaemia cell line EoL-1 (Sugimura et al, 1999). In a recent study, troglitazone induced G1 arrest in pancreatic carcinoma cell line PK-1 was found to be associated with increased p27 Kipl but unchanged p21 Cipl/Wafl and p18 Ink4c levels . Cell cycle withdrawal during PPARγ induced adipogenesis was also found associated with an initial increase in both p27 Kipl and p21 Cipl/Wafl , followed by a decline in p21 Cipl/Wafl , increase in p18 Ink4c with sustained p27 Kipl level (Morrison and Farmer, 1999). Difference in the extent of G1 or G2 cell cycle arrest in Hep G2 cells with troglitazone, pioglitazone or 15d-PGJ 2 may be due to differences in cascade expression of the CDK inhibitors. However, CDK inhibitors other than p27 Kipl , p21 Cipl/Wafl and p18 Ink4c may also be involved in the PPARγ ligands induced cell cycle arrest. Possibility of an initial increased p21 Cipl/Wafl followed by declined level in troglitazone treated Hep G2 cells as reported in coupling growth arrest and adipocyte differentiation (Morrison and Farmer, 1999) can not be excluded until further time course studies are done. Further studies are also required to reveal whether inhibition of the p27 Kipl or p18 Ink4c increase could inhibit the growth arrest in Hep G2 as found in PK-1 cells . PPARγ ligands have been shown to drive differentiation process in various malignant cells (Tontonoz et al, 1997;Kubota et al, 1998;Sarraf et al, 1998;Chang and Szabo, 2000). When we investigated the expression of AFP and albumin in Hep G2 cells after treatment with troglitazone, we found that there was a dosedependent decrease in the AFP expression. The expression of albumin showed a little increase with 5 and 10 µM troglitazone but decrease with 25 µM. The decreased albumin expression at 25 µM may be due to associated cytotoxic response of troglitazone on Hep G2 cells. However, the inhibition of AFP expression was more prominent over albumin inhibition and results of repeated Northern analysis revealed that instead of variation in the amount of decrease in AFP expression or increase in albumin expression, the dose-dependent increase in ratio of albumin and AFP expression remained consistent.
Hepatocellular carcinoma is one of the most lethal malignancies where chemoprevention is recommended and different type newer agents are already tried on human subjects (Oka et al, 1995;Muto et al, 1996;Jacobson et al, 1997). PPARγ ligands exhibited a marked growth inhibitory potential on hepatocellular carcinoma cells through induction of G1 cell cycle arrest and recent studies have also shown that activation of PPARγ can inhibit the profibrogenic and proinflammatory actions of hepatic stellate cells (Galli et al, 2000;Marra et al, 2000;Miyahara et al, 2000). Taken together, we conclude that PPARγ ligands may also prove beneficial for primary or secondary chemoprevention of hepatocellular carcinoma. However, further studies will be necessary to evaluate their safety and therapeutic potential.