Ligand activation of peroxisome proliferator-activated receptor γ (PPARγ) results in the inhibition of proliferation of various cancer cells. The aim of this study is to investigate the mechanisms of cell growth inhibition of hepatocellular carcinoma (HCC) cell lines by the PPARγ ligand, troglitazone.
Six HCC cell lines were used to study the effects of troglitazone on cell growth by 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) assay, on cell cycle by flow cytometry, and on the cell cycle-regulating factors of late G1 phase by Western blotting. Apoptosis assays were performed by flow cytometry using membrane, nuclear, cytoplasmic, and mitochondrial markers. Caspase inhibitors were used to analyze the mechanisms of apoptosis induced by troglitazone.
Troglitazone showed a potent dose-dependent effect on the growth inhibition of all six HCC cell lines, which were suppressed to under 50% of control at the concentration of 10 μmol/L. The growth inhibition was linked to the G1 phase cell cycle arrest through the up-expression of the cyclin-dependent kinase inhibitors, p21 and p27 proteins, and the hypophosphorylation of retinoblastoma protein. Troglitazone induced apoptosis by caspase-dependent (mitchondrial transmembrane potential decrease, cleavage of poly [adenosine diphosphate ribose] polymerase, 7A6 antigen exposure, Bcl-2 decrease, and activation of caspase 3) and caspase-independent (phosphatidylserine externalization) mechanisms.
Hepatocellular carcinoma (HCC) is one of the most common malignancies worldwide.1 An effective treatment for HCC patients is urgently needed. Hepatitis B virus and hepatitis C virus are associated with HCC in many cases. However, little is known about the mechanisms of hepatocarcinogenesis.
Peroxisome proliferator-activated receptors (PPAR) are members of a nuclear receptor superfamily. The isoform, PPARγ, acts as a ligand-sensitive transcription factor.2, 3 Activated PPARγ heterodimerized with retinoid X receptor binds to peroxisome proliferator-responsive elements and effects the transcription of target genes.4 PPARγ was first reported to play a role in the differentiation of adipocytes.3 Thiazolidinediones (troglitazone, pioglitazone, and rosiglitazone), a new class of antidiabetic agents, are synthetic ligands for PPARγ. Although their precise mechanism of action is unknown, thiazolidinediones improve the sensibility to insulin action, probably mediated via activation of PPARγ.5
Recent studies showed that troglitazone also causes growth inhibition by inducing the differentiation or apoptosis of various human malignant cells, including myeloid leukemic cells,6 breast carcinoma,7, 8 prostatic carcinoma,9 liposarcoma,10 endothelial cells,11 normal and malignant B-lineage cells,12 nonsmall cell lung carcinoma,13 colorectal carcinoma,14, 15 gastric carcinoma,16 pancreatic carcinoma,17, 18 esophageal adenocarcinoma,19 and HCC.20, 21 However, some studies indicated that troglitazone promoted mouse colon polyps or tumors22, 23 and hemangiosarcoma in the B6C3F1 mouse line,24 showing that the value of PPARγ ligands remains an open question. The mechanism by which troglitazone inhibits the growth or differentiation of tumor cells is still unknown.
Proliferation of cells is regulated by cyclin/cyclin-dependent kinase (CDK) complexes and CDK inhibitors. The retinoblastoma (Rb) gene product plays a key role in the cell cycle. Phosphorylation of Rb protein by cyclin/CDK complexes leads cells from G1 to S phase. The CDK inhibitors, p21 and p27, blocked the phosphorylation of Rb protein by inhibiting cyclin/CDK inhibitors, resulting in the G1 arrest of cells. In previous reports, growth inhibition of HCC cell lines was induced by troglitazone via the up-regulation of p21 or p27.20, 21
Apoptosis is a tightly regulated form of physiologic cell death that is dependent on the expression of cell-intrinsic suicide machinery. The death signals are transmitted from various receptors, eventually resulting in the activation of a proteolytic system composed of a cascade of proteases called caspases. Mitochondria are another important component of the programmed cell death pathway, which is intimately entwined in numerous death pathways. Members of the Bcl-2 protein family help to regulate the mitchondrial activity, being “arbiters of cell survival.”25
In this study, we tested the hypothesis that troglitazone inhibited the growth of human HCC cell lines through both a restriction point control of the late G1 phase of the cell cycle and induction of programmed cell death. We explored the membrane, nuclear, and mitochondrial pathways of programmed cell death induced by troglitazone in human HCC cell lines.
MATERIALS AND METHODS
Troglitazone and rosiglitazone were provided by Sankyo (Tokyo, Japan) and pioglitazone was provided by Takeda Pharmaceutical (Osaka, Japan). The stock solution was prepared at a concentration of 100 mmol/L in dimethyl sulfoxide (DMSO). The final concentration of DMSO in the culture medium did not exceed 0.1%.
Effects of Troglitazone, Pioglitazone, and Rosiglitazone on the Growth of HCC Cell Lines
Five established HCC cell lines (SK-Hep 1, Hep 3B, Hep G2, PLC/PRF/5, and Mahlavue) and SUHC-1, which was established from the HCC tissue of an HCV-infected patient in our laboratory,26 were used in this study. To study the effects of these chemicals on cell growth, various doses (0.01, 0.1, 1, 10, and 100 μmol/L) or vehicle (control) were added to each cell culture in 96-well flat plates on the second day of the experiment. We assayed the viable cell number using the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5- diphenyltetrazolium bromide; Sigma, St. Louis, MO) methods as described previously.18 MTT absorbance was determined on days when cells with vehicle came close to confluence. Eight wells were used for each treatment and experiments were repeated twice. Cytotoxicity of each ligand was examined by the trypan blue exclusion test.
Cell Cycle Analysis
Cells were plated and allowed to grow to 30% confluence, then treated with vehicle or troglitazone (10 μmol/L) for 3 days. Cells were harvested and stained with a DNA staining solution containing propidium iodide (PI; 50 μg/ml) and RNase (1.8 U/mL). The cells were analyzed with a FACScan flow cytometer equipped with an argon laser (Becton Dickinson Immunocytometry System, Mountain View, CA).
Immunoblotting was performed as described previously.18 Briefly, the cells were homogenized using an ultrasonic cell breaker (Poweresonic model 50, Yamato Kagaku, Tokyo, Japan) and lysed in cell lysis buffer. Extracts equivalent to 50 μg of total protein were separated by sodium dodecyl sulfate-polyacrylamide gel and transferred to supported nitrocellulose membrains (Gibco BRL, Gaithersburg, MD) with a plate electrode apparatus (Idea Scientific, Minneapolis, MN). The filters were blocked with 5% nonfat milk, after which they were incubated with each antibody and with horseradish peroxidase-conjugated rabbit antimouse or donkey antirabbit Ig (1:1000; Amersham, Arlington Heights, IL) in tris-buffered saline with Tween-20 (Sigma). Bound antibody was detected with an enhanced chemiluminescence system (ECL kit; Amersham Pharmacia Biotech, Buckinghamshire, UK) and exposed to X-ray films. To verify that equal amounts of total proteins were present in all lanes, β-actin staining served as the internal standard for all filters. The experiments were repeated twice. The following primary antibodies were used in this study: rabbit polyclonal antibody to p27 (Santa Cruz Biotechnology, Santa Cruz, CA); mouse monoclonal antibodies (MoAbs) to PPARγ (Santa Cruz Biotechnology), p21 (Transduction Laboratories, Lexington, KY), RB protein (PharMingen, San Diego, CA), and β-actin (Sigma).
The annexin V fluorescent assay was performed as recommended by the manufacturer, the cells being extracted from the culture media, washed twice in phosphate-buffered saline (PBS) and stained with 2 μL annexin V-Fluos fluorescein isothiocyanate (FITC)-labeled (Boehringer Mannheim, GmbH, Mannheim, Germany) in the incubation buffer (10 mmol/L Hepes/NaOH, pH 7.4, 140 mmol/L NaCl, 5 mmol/L CaCl2). For discrimination between early apoptotic cells and cells with permeabilized membranes (late apoptotic or secondary necrotic cells), a counterstain of 5 μg/mL PI (Calbiochem, La Jolla, CA) was used.
The poly (adenosine diphosphate [ADP]-ribose) polymerase (PARP) assay was performed using the FITC-conjugated anti-PARP cleavage site (214/215) antibody from BioSource International (Camarillo, CA). The cells were fixed and permeabilized using the IntraStain kit (Dako A/S, Glostrup, Denmark), stained with the anti-PARP antibody, and analyzed for the green fluorescence by flow cytometry.
The mitochondrial assays were performed using 3,3'-dihexylocarbocyanine iodide (DiOC), the Apo 2.7 MoAb, and the anti–Bcl-2 MoAb. DiOC is a mitochondrial transmembrane potentiometric fluorescent marker. A decrease in its fluorescence reveals the disruption of the mitochondrial transmembrane potential, an early and irreversible step of ongoing apoptosis.27 The cells were incubated with 40 nM DiOC (Molecular Probes, Eugene, OR) in PBS for 20 minutes at 37 °C, washed twice, and analyzed by flow cytometry.
The Apo 2.7 MoAb reacts with a 38-kilodalton mitochondrial membrane protein (the 7A6 antigen), which is exposed on cells undergoing programmed cell death. Its expression represents an early event of apoptosis rather than a final product of dead cells.28 The cells were fixed and permeabilized using the IntraStain kit, incubated for 15 minutes at room temperature with the phycoerythrin (PE)-conjugated Apo2.7 antibody (Immunotech, Marseille, France), washed, and subjected to flow cytometric analysis.
For the Bcl-2 assay, the cells were fixed and permeabilized using the two-step IntraStain kit and incubated with the FITC-conjugated mouse anti–Bcl-2 MoAb (clone 124, Dako). The subsequent flow cytometric analysis quantified the mean fluorescence intensity of the bound fluorescent antibody, which was directly proportional to intracytoplasmic Bcl-2 protein levels.
The caspase-3 fluorescent assay was performed by fixing and permeabilizing the cells with the IntraStain kit and staining with the PE-conjugated antiactive caspase-3 antibody (PharMingen International, Tokyo, Japan), which recognizes active human caspase-3, as opposed to pro-caspase-3. The following flow cytometric analysis discriminated the cells that have or are not exposed to the conformational epitope formed by the cleavage of pro-caspase-3.
The inhibitory assays were performed using the nonselective, cell-permeable inhibitor of ICE-like and CPP32-like caspases, N-benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl ketone (Z-VAD-FMK), and the reversible inhibitor of caspase-3, caspase-1, and caspase-7, i.e., N-acetyl-Asp-Glu-Val-Asp-al (Ac-DEVD-CHO). Both were provided by Sigma. The cells were cultured in the presence of troglitazone with or without adding 30 μmol/L of Z-VAD-FMK or 20 μmol/L Ac-DEVD-CHO. The differences in programmed cell death induction were assayed by the methods described above.
All flow cytometry experiments were performed in triplicate. The paired Student t test was used to determine the statistical significance of the data obtained and to compare the means of the two groups. StatView software was used (Abacus Concepts, Berkeley, CA). A P value of less than 0.05 represented a statistically significant difference between the values of two group means. Kolmogorov–Smirnov statistics were used to assess the flow cytometric histograms and the significance of their shifting.
Effects of Troglitazone, Pioglitazone, and Rosiglitazone on the Proliferation of Each Cell Line
Troglitazone showed a potent dose-response effect on the growth of HCC cell lines. Thirty to 60% of growth inhibition was observed at 10 μmol/L. Pioglitazone had a weaker effect on the HCC cell lines than troglitazone. Rosiglitazone showed an antiproliferative effect only at a concentration of 100 μmol/L on all six HCC cell lines (Fig. 1). All six HCC cells expressed various levels of PPARγ protein (data not shown). We found no relationship between the sensitivity to troglitazone and the intensity of PPARγ protein expression in each cell line.
Cell Cycle Analysis
Representative results of cell cycle analysis in the SK-Hep 1 and Hep 3B cell lines are shown in Figure 2. The substantial increase in the fraction of cells in the G1 phase (51% and 53%, respectively) was observed after treatment for 3 days with 10 μmol/L troglitazone compared with the G1 fraction of control groups (73% and 64%, respectively), indicating that troglitazone induced the G1 phase arrest of the cell cycle.
Analysis of G1 Restriction Point-Regulating Agents
Western blotting analysis of the representative cell line, SK-Hep 1, showed that troglitazone induced a substantial increase in the p21 and p27 proteins in a time and dose-dependent manner after treatment with 30 μmol/L troglitazone or vehicle (control) for indicated times (12 and 24 hours) and at various concentrations (10, 20, and 30 μmol/L) for 24 hours. Rb protein was decreased and the hypophosphorylated forms of Rb protein were increased by troglitazone (Fig. 3).
Troglitazone Induces Programmed Cell Death in the HCC Cell Lines
After 36 hours of incubation with troglitazone concentrations of 30 μmol/L and higher, apoptosis was significantly induced in the SK-Hep 1 and Hep 3B HCC cell lines (Fig. 4). Apoptosis reached near-maximum values at the troglitazone concentration of 50 μmol/L (5.6 ± 1.5% control vs. 60.6 ± 19.5% at 50 μmol/L troglitazone for the SK-Hep 1 cell line and 9.1 ± 3.4% control vs. 58.6 ± 19.7% at 50 μmol/L troglitazone for the Hep 3B cell line). The two cell lines displayed comparable levels of programmed cell death induction (Fig. 4A). As shown in Figure 4B, programmed cell death induction occurred between 12 and 24 hours. Near-maximum levels of apoptosis occurred at 24 hours, as determined by phosphatidylserine externalization on the outer cellular membrane.
Apoptosis Induction by Troglitazone Is a Phenomenon with a Caspase-Dependent and Caspase-Independent Component
Programmed cell death induction was measured by the fluorescent annexin V method, which assesses phosphatidylserine externalization on the outer cellular membrane, the intracytoplasmic appearance of active caspase-3, and the cleavage of PARP (Fig. 5). By using the caspase inhibitors, Z-VAD-FMK and Ac-DEVD-CHO, both caspase-3 activation (Fig. 5B) and PARP cleavage (Fig. 5C) in the troglitazone-treated samples were completely blocked. However, the phosphatidylserine externalization on the outer cellular membrane was not blocked (Fig. 5A), demonstrating that programmed cell death induction by troglitazone has caspase-dependent and independent components.
Mitochondrial Involvement in Troglitazone-Induced Programmed Cell Death of the HCC Cell Lines
Mitochondrial transmebrane potential, ΔΨm, as measured by the potentiometric marker, DiOC, was significantly decreased in cells undergoing apoptosis after troglitazone treatment, demonstrating the existence of mitochondrial potential transition in the troglitazone-treated cells (Fig. 6). The appearance of ΔΨmlow cells was completely blocked by the two caspase inhibitors, Z-VAD-FMK and Ac-DEVD-CHO, indicating that this phenomenon was also under the control of the activated caspases.
Another mitochondrial marker of programmed cell death induction, the appearance of the 7A6 antigen detected by the Apo 2.7 MoAb, was also completely inhibited by Z-VAD-FMK and Ac-DEVD-CHO (Fig. 6B), demonstrating that this is also a caspase-dependent phenomenon.
Finally, the Bcl-2 oncoprotein, which is an integral inner mitochondrial membrane protein that blocks programmed cell death, was also significantly decreased in the troglitazone-treated samples. This down-regulation was inhibited by the caspase inhibitors, Z-VAD-FMK and Ac-DEVD-CHO (Fig. 6C), demonstrating that this Bcl-2 decrease was also under the control of the activated caspases.
In this study, we demonstrated that the PPARγ ligand, troglitazone, showed a dose-dependent growth inhibition effect on HCC cell lines, which is consistent with observations on other HCC cell lines.20, 21 Conversely, pioglitazone and rosiglitazone had weaker effects on the HCC cell lines compared with troglitazone. These three thiazolidinediones are similar in their effects on blood glucose. All class members demonstrated effective glycemic control, both as monotherapy and in combination with sulphonylureas, metformin, or exogenous insulin. The glycemic control was probably mediated by the activation of PPARγ, although the precise mechanism is unknown.5 The difference of effects on tumor cell growth among these three members should be investigated.
We showed by flow cytometric cell cycle analysis that the growth-inhibitory effect of troglitazone was related to G1 phase cell cycle arrest, suggesting the contribution of cell cycle regulating factors that act at a restriction point of late G1. Kawa et al.18 reported the G1 cell cycle arrest of pancreatic carcinoma cell lines by troglitazone. Koga et al.20 and Rumi et al.21 showed similar results using five and four HCC cell lines, respectively, all of which except Hep G2 were different from our cell lines. We demonstrated that the expression levels of the CDK inhibitors, p21 and p27, were significantly increased and the phosphorylated Rb protein was markedly decreased in a time and dose-dependent manner by troglitazone in SK-Hep 1 cells. Koga et al. showed that the p21 protein was up-regulated in four of five cell lines and that the p27 and p18 proteins were up-regulated in one of five cell lines. Conversely, Rumi et al. showed that the p27 and p18 proteins were increased but that p21 was decreased by troglitazone in Hep G2 cells. Kawa et al.18 reported that troglitazone induced up-regulation of p21 but not p27 protein expression in pancreatic carcinoma cell lines. Motomura et al.,17 however, demonstrated that troglitazone increased p27 but not p21 or p18 protein levels in another pancreatic carcinoma cell line. These results indicate that there are several pathways of G1 phase cell cycle arrest among HCC and other cancer cell lines.
Apoptosis induction by PPARγ ligands was demonstrated in a variety of cell lines.8, 12, 13, 16, 18, 19 In the HL-60 promyelocytic/myeloblastic leukemia cell line, troglitazone induced DNA fragmentation. The morphologic changes of apoptosis (shrinking with nuclear condensation and fragmentation) at concentrations of 25–50 μmol/L29 were similar to those used in our experiments.
To our knowledge, this is the first study to explore the mechanisms of apoptosis induced by troglitazone in HCC cell lines. Of importance for defining the pathway of troglitazone-induced programmed cell death is the finding that phosphatidylserine externalization (as a membrane early event during apoptosis induction, measured by annexin V binding) is a caspase-independent phenomenon (which cannot be blocked by the caspase inhibitors, Z-VAD-FMK and Ac-DEVD-CHO). PARP cleavage (representative of the nuclear features of apoptosis) and the mitochondrial markers of programmed cell death (mitochondrial transmembrane potential decrease measured by DiOC levels, 7A6 antigen exposure measured by the Apo 2.7 MoAb, and the decrease of Bcl-2 levels) were caspase-dependent phenomena. These results partially support the data of Pettitt and Cawley.30 They reported fludarabine-induced apoptosis in chronic lymphocytic leukemia cells, where phosphatidylserine exposure was a caspase-independent phenomenon. However, in the case of fludarabine, mitochondrial depolarization was also reported to be caspase independent. For troglitazone, mitochondrial depolarization was totally blocked by the caspase inhibitors, Z-VAD-FMK and Ac-DEVD-CHO (Fig. 6A). The apoptosis assays used in the current study investigated different cellular compartments (e.g., annexin V, the membrane; PARP, the nucleus; caspase-3, the cytoplasm; DiOC, Apo 2.7; and Bcl-2, the mitochondria). However, these assays addressed the early events taking place during programmed cell death induction, rather than the late events represented by DNA fragmentation and characteristic morphologic changes.
Our results indicate significant apoptosis induction by troglitazone in HCC cell lines, which is at variance with the data reported by Koga et al.20 They did not obtain a significant increase in the sub-G1 cell population (representing apoptotic cells) after exposure to 50 μmol/L troglitazone, even if the growth-inhibitory effect was found in all lines. However, their HCC cell lines (HLF, HuH-7, HAK-1A, HAK-1B, and HAK-5) were different from those used in this study and the only method used to investigate programmed cell death induction was the PI staining used to analyze the cell cycle. The apoptosis assays used in our study detected the earliest events of programmed cell death induction (phosphatidylserine externalization, caspase-3 activation, and the decrease of mitochondrial transmembrane potential), as opposed to the appearance of the sub-G0/G1 hypodiploid population of cells, a very late event in the chronology of programmed cell death induction. This apparent discrepancy of findings between the two studies may be explained by the different cell lines used, the different methods employed, and the studied phenomena occurring at different time points. In their study, Koga et al.20 described ultrastructural alterations in troglitazone-treated HCC cells, with electron-dense bodies in the cytoplasm being located adjacent to mitochondria. Some of these electron-dense bodies were derived from a part of the mitochondrial membrane, often demonstrating a myelin-like structure suggestive of degraded membrane constituents. The intracytoplasmic electron-dense bodies were found not only in the most troglitazone-sensitive HCC cell line, but also in the least sensitive one, suggesting that mitochondrial involvement was ubiquitous. These findings support our results and complement our data regarding the mitochondrial involvement in the process of programmed cell death induction by troglitazone. Three different assays point to the mitochondria as an important component of apoptosis initiation, i.e., disruption of the mitochondrial transmembrane potential, 7A6 antigen exposure, and the decrease of Bcl-2 levels.
The observation that troglitazone induces programmed cell death in the HCC cell lines studied may offer an explanation for the clinical reported cases of fulminant hepatitis after administration of the drug,31–33 which determined the withdrawal of troglitazone from clinical use. The therapeutic implications of the troglitazone growth-inhibitory effect on HCC still need to be studied. Further research is necessary to define the molecular mechanisms of apoptosis induction and to establish possible HCC approaches using troglitazone or modified analogs of the thiazolidinedione class of drugs.